THE DMC FRAMEWORK APPLIED TO WORKING MEMORY

Get Complete Project Material File(s) Now! »

Participants.

A sample of 62 participants completed the experiment (15 males and 47 females; age ranging from 17 to 25, M = 20.58, SD = 1.99). All participants were undergraduate students from the University of Savoy participating for course credit. The inclusion criteria were identical to Experiment 1a. None of the participants had participated in Experiment 1a.

Materials.

Working memory task. Working memory capacity was measured with the CCS (see Appendix A). The dependent variable on the task was the composite working memory score.
Prospective memory task. The main ball-catching task was identical to Experiment 1a, with the following exceptions. Instead of receiving instructions concerning the red balls, participants were instructed to press the spacebar every two minutes during the ball-catching task. A correct hit was scored whenever the participant pressed the spacebar within a time window of five seconds (i.e., 2500ms) around the expected time (this procedure was similar to Kliegel, Martin, McDaniel, & Einstein, 2001). Participants were allowed to check the time during the experiment: pressing the A key displayed a clock indicating the amount of time that had elapsed since the beginning of the task. To prevent participants from keeping the clock consistently displayed throughout the task, it only remained on-screen for one second after pressing the A key and could not be displayed again for the next three seconds (see e.g. Kliegel et al., 2001, for a similar procedure). Two dependent variables were collected in the task: the total ball-catching score was recorded to index performance in the main task, and the percentage of times when the spacebar was correctly pressed served to index prospective memory performance.
As in Experiment 1a, we expected the correlation between working memory capacity and prospective memory performance to be modulated by the difficulty of the ball-catching task, as operationalized by the size of the paddle. However, manipulating difficulty in the same way as Experiment 1a – by having participants perform successive periods of easy and difficult conditions of the ball-catching task – was not an option: the scheduled alternation of phases would have provided participants with an external time cue, which could have interfered with the time-based prospective memory task. As a consequence, the difficulty of the ball-catching task was manipulated as a between-subjects variable in this experiment. Thus, the task only included a single continuous period of 15 minutes with constant difficulty rather than alternating periods of 150 seconds as in Experiment 1a. In the final sample, 29 participants completed the easy condition and 33 participants completed the difficult condition.
Time estimation task. The time estimation task simply had participants press the spacebar every 2 minutes for 10 minutes. In order to prevent participants from focusing on the time estimation task and implementing strategies such as counting the seconds, they also completed a focal task during these 10 minutes. The focal task was the Mesulam continuous performance test (Mesulam, 1985), a simple symbol cancellation task. The task consisted of paper sheets with various printed symbols (such as uppercase letters); participants had to cross out all symbols of a given type on the sheet (for example, all instances of the letter A). This test was chosen to demand as little attention as possible so as not to bias time estimation performance. The dependent variable in this task was the median number of seconds between the moment when the participant should have pressed the spacebar and the moment when he actually pressed the spacebar.

Procedure.

The procedure was similar to Experiment 1a with the following exceptions. Participants completed the prospective memory task, the time estimation task and the CCS, in order. The whole procedure took approximately 45 minutes. The PRMQ was not included in Experiment 1b.

Results

Preliminary analyses.

Working memory scores were normally distributed and close to the population average (M = -0.03, SD = 0.71, skewness = 0.10, kurtosis = -0.59). One participant was excluded because of his performance in the processing tasks, yielding a total sample of 61 subjects.
A series of analyses was performed to check the correct functioning of the experimental paradigm. Performance in the ball-catching task was higher in the easy condition (M = 467.43, SD = 229.20) than in the difficult condition (M = 86.06, SD = 184.80); this difference was significant, F(1, 59) = 51.76, MSE = 42563, p < .001,²p = .47. As in Experiment 1a, performance in the ball-catching task was positively correlated with working memory capacity, F(1, 57) = 9.33, MSE = 37778, p = .003,²p = .14, r = .27, but this correlation did not depend on task difficulty, F(1, 57) = 0.41, MSE = 3778, p = .527, ²p = .01.
On average, participants correctly performed the prospective action 32% of the time (SD = 23, range = 0 – 67). Contrary to Experiment 1a, performance in the prospective memory task was comparable in the easy condition (M = 34, SD = 23) and in the difficult condition (M = 30, SD = 24) of the ball-catching task, F(1, 59) = 0.30, MSE = 551.55, p = .583,²p = .01. A total of 8 participants (13% of the sample) entirely disregarded the prospective memory task instructions and never pressed the spacebar throughout the task; as in Experiment 1a, a logistic regression indicated that the probability of forgetting the prospective instructions was not correlated with working memory capacity,²(1) = 0.00, p = .955. All participants who completely forgot the prospective memory task were excluded from analyses on prospective performance (see Experiment 1a, p. 105).
Scores on the time estimation task were normally distributed, (M = -0.47, SD = 27.40, skewness = -0.23, kurtosis = 0.69). As expected, performance in the time estimation task was marginally correlated with performance in the prospective memory task, F(1, 58) = 3.02, MSE = 535.29, p = .087,²p = .05, r = .22; importantly, however, working memory capacity did not correlate with time estimation performance, F(1, 56) = 0.88, MSE = 677.02, p = .353, ²p = .02, r = .12.

Main analyses.

Our first hypothesis was that working memory capacity would be positively correlated with prospective memory performance. The correlation was significant, F(1, 49) = 9.42, MSE = 352.44, p = .003,²p = .16, r = .40 (see Figure 9), congruent with our expectations. The correlation was still significant when controlling for performance in the ball-catching task, F(1, 45) = 6.98, MSE = 323.18, p = .011,²p = .13, r = .55, and when controlling for performance in the time estimation task8, F(1, 44) = 8.24, MSE = 334.76, p = .006,²p = .16, r = .39.
Our second hypothesis was that the correlation between working memory capacity and prospective memory performance would be modulated by task difficulty; we expected the correlation to be lower in the difficult condition of the ball-catching task. Contrary to this hypothesis, the two-way interaction was not significant, F(1, 47) = 0.01, MSE = 360.73, p = .920,²p = .00; the correlation between working memory capacity and performance was similar for participants who completed the easy version (r = .46) and participants who completed the difficult version of the ball-catching task (r = .37).
Figure 9. Correlation between working memory capacity and prospective performance.

Discussion

In Experiment 1a, we did not observe the expected correlation between working memory capacity and performance in an event-based prospective memory task. We hypothesized that this null result was due to the nature of the prospective memory task, in which the salient event cue promoted the use of spontaneous retrieval mechanisms – or reactive control – in all participants. In Experiment 1b, we tested the same relationship with a time-based prospective memory task, designed to require preparatory processes and prevent the use of spontaneous retrieval mechanisms. This time, working memory capacity was related to performance in the prospective memory task; this relationship was not mediated by performance in the main task or by the time estimation abilities of participants. Thus, our results support the hypothesis that participants with high working memory capacity have a higher tendency to use proactive control, which elicits a higher performance in tasks where preparatory processes play a central role.
Interestingly, this conclusion allows for an elegant interpretation of results in the prospective memory literature. Recall that certain studies have observed a correlation between working memory capacity and prospective memory performance, while others observed no relationship (see pp. 99-100). Our results suggest that the correlation between working memory capacity and prospective memory performance depends on the degree to which the prospective memory task relies on preparatory processes. In other words, a prospective memory task would correlate with working memory capacity if the task cannot be solved with spontaneous retrieval, for example because there is no salient cue event to trigger the retrieval – which especially includes time-based prospective tasks. Conversely, an event-based prospective task including salient cue events would not correlate with working memory capacity. Of course, event-based prospective memory task with high working memory demands could also correlate with working memory capacity (which includes event-related paradigms requiring participants to detect one of six targets) Overall, the literature seems to fit this interpretation. A single study tested the relationship between working memory capacity and time-based prospective memory and observed a significant correlation between the two variables (Kretschmer et al., 2013). One study observed a correlation between working memory capacity and event-based prospective memory performance, but only when participants had to wait for several dozens of seconds after they had detected the cue event to perform the prospective action; such a delay certainly required preparatory processes since it made the prospective action clearly separate from the cue that could have elicited spontaneous retrieval (Einstein et al., 2000). Other studies used prospective memory tasks associated with low working memory demands, but in which the cue event was not easily detected, which presumably makes spontaneous retrieval much less relevant and emphasizes the role of preparatory processes; these studies also observed a correlation with working memory capacity (Brewer, Knight, Marsh, & Unsworth, 2010).
The studies which used prospective memory tasks associated with high working memory demands also found a correlation with working memory capacity (R. E. Smith, 2003; R. E. Smith & Bayen, 2005; R. E. Smith, Persyn, & Butler, 2011). Conversely, the studies which used event-based prospective memory tasks with easily detected cue events, without working memory demands, and without an imposed delay between the cue and the prospective action to be performed reported a weak or non-existent correlation between prospective memory performance and working memory capacity. The latter finding is consistent with the idea that all participants resorted to spontaneous retrieval in these situations, as in our own Experiment 1a (Cherry & LeCompte, 1999; Breneiser & McDaniel, 2006).

Table of contents :

Part 1: Introduction
PREAMBLE
CHAPTER 1. WORKING MEMORY AND HIGH-LEVEL COGNITION
CHAPTER 2. COGNITIVE CONTROL AS AN EXPLANATORY FACTOR
CHAPTER 3. THE DMC FRAMEWORK APPLIED TO WORKING MEMORY
Part 2: Experimental section
CHAPTER 4. PROACTIVE CONTROL AS PREPARATORY PROCESSING
CHAPTER 5. PROACTIVE CONTROL AS SENSITIVITY TO CONTEXT
CHAPTER 6. PROACTIVE CONTROL MEASURED IN CLASSIC TASKS
CHAPTER 7. BRAIN ACTIVITY AS A PROACTIVE CONTROL INDEX
CHAPTER 8. THE EXPLANATORY VALUE OF PROACTIVE CONTROL
Part 3: General discussion
CHAPTER 9. GENERAL DISCUSSION
Part 4: Appendices
REFERENCES
APPENDICES A-G
TABLE OF CONTENTS

GET THE COMPLETE PROJECT