Elsevier

Acta Psychologica

Volume 125, Issue 3, July 2007, Pages 319-333
Acta Psychologica

Cross-task repetition amnesia: Impaired recall of RSVP targets held in memory for a secondary task

https://doi.org/10.1016/j.actpsy.2006.08.006Get rights and content

Abstract

People often fail to select and encode the second of two targets presented within less than 500 ms in rapid serial visual presentation (RSVP), an effect known as the attentional blink. We investigated how report of the two targets is affected when one of them is maintained in working memory for a secondary, memory-search task. The results showed that report of either target was impaired when it was a member of the memory set relative to when it was not. This effect was independent of both the temporal interval separating the RSVP target from the presentation of the memory set and the interval separating the targets. We propose that the deficit in recall occurs because the association between a target and the memory-search task interferes with the formation of a new association between that target and the following RSVP task, with the result that observers may be biased to ascribe the target only to the memory set.

Introduction

Working memory and selective attention are core cognitive functions that enable us to interact with our environment in a goal-directed fashion. Baddeley (1997) defined working memory as “a system for temporarily holding and manipulating information as part of a wide range of essential cognitive tasks such as learning, reasoning and comprehending” (p. 49). Selective attention, on the other hand, controls the input of perceived information into the working memory system (e.g., Baddeley, 1986, Broadbent, 1958, Reeves and Sperling, 1986). Thus, in order to update goals and perceptions with new and relevant information from the world, processes of selective attention and working memory must interact.

The interaction between information held in memory and attentional selection can be observed in many common tasks, such as searching for a particular object in a field of other objects. In the context of search tasks, Desimone and Duncan (1995) proposed that attentional selection is driven by short-term descriptions (i.e., templates) of objects held in working memory. One hypothesis that follows from this view is that attentional selection of visual stimuli should be biased towards those stimuli that correspond to representations held active in working memory. Consistent with this hypothesis, Downing (2000) found that reaction time to a probe presented at the location of a face that had previously been memorized was faster than reaction time to probes presented at the locations of “new” faces, suggesting that faces held in memory more readily attracted attention than new, non-memorized, faces. Further support for memory-contingent attentional capture was reported by Pashler and Shiu (1999), who found that a previously imagined object automatically attracted attention when it was presented in a sequence of other objects. Specifically, when a search target (a digit) was presented shortly after the imagined object, target identification was hindered – presumably because of the time required to disengage attention attracted to the previously imagined object. Taken together, these findings indicate that objects held in visual working memory automatically attract attention when they are encountered in the visual environment.

In the present study, we investigated whether maintaining letters in working memory enhances selection and encoding of these letters when they are used as targets in a dual-target RSVP task (Broadbent & Broadbent, 1987). In this task, observers are to identify two visual targets (e.g., letters) embedded in a sequence of distractors (e.g., digits). Previous studies using this paradigm have shown that recall of the second target (T2) is often impaired when it is presented within 200–500 ms of the first target (T1). This phenomenon, called the attentional blink (AB; Raymond, Shapiro, & Arnell, 1992), has been shown to reflect a temporal limitation in our ability to select successive visual stimuli (Nieuwenstein et al., 2005, Nieuwenstein and Potter, 2006). In particular, Nieuwenstein et al., showed that the deficit in T2 report is substantially reduced when this target is directly preceded by a distractor that shares one of two target defining features (e.g., color). This result suggests that the AB reflects a failure to allocate attention to T2 in time before it is replaced by its mask, a failure that can be overcome by precuing the target with a distractor that shares a feature with the target.

Given that the AB effect depends critically on the allocation of attention to T2, we hypothesized that report of T2 should be facilitated if the target were to capture attention because it is held in memory. To test this hypothesis, we had observers perform an AB task during the retention interval of a memory-search task that required them to maintain a set of letters in memory for a later recognition task. The AB task required identification of two letter targets presented in an RSVP sequence of digit distractors, so the identity of the targets was not known in advance. On half the trials, T2 (Experiments 1A and 1B) or T1 (Experiment 2) was included in this memory set. In order to control for the effect of maintaining a memory set, we also included conditions in which the AB task was performed in isolation. The effect of a match between one of the letters in the memory set and the RSVP target was determined by comparing identification performance for matching and non-matching targets, and the effect of maintaining a memory load was determined by comparing identification performance for the condition in which the memory set did not include either of the targets with performance in a condition in which the AB task was performed in isolation.

Section snippets

Participants

A total of 30 undergraduate students from Utrecht University participated in the experiments. Fourteen students participated in Experiment 1A (all stimuli in uppercase), and 16 in Experiment 1B (memory-set items in lowercase and targets in uppercase). All participants gave informed consent and received payment of €10 for participating.

Apparatus and stimuli

The stimuli were presented on a 17-in. monitor with a 75-Hz refresh rate. The generation of stimuli and collection of responses were controlled using E-Prime 1.0

Results Experiments 1A and 1B

The memory probe was correctly classified on 95% of the trials in which a memory set was presented. On these trials, T1 was identified correctly in 91% of the cases. Fig. 2a shows T1 performance across lag and memory-load condition. The ANOVA of the effects of lag (1, 3, 5, or 8) and memory-load condition (no load, T2 not in set or T2 in set) showed that both factors had a significant effect on T1 report, F(3, 87) = 40.3, p < .001, MSE = .003, and F(2, 58) = 8.7, p = .001, MSE = .007, respectively, while the

Discussion Experiments 1A and 1B

The results from Experiments 1A and 1B reveal a counterintuitive finding: Report accuracy for the second of two RSVP targets was impaired rather than improved when the corresponding letter was included in a memory set that was retained throughout the presentation of the RSVP trial. This impairment for targets retained in memory for the memory search task suggests that there was item-specific cross-talk between the two task, an effect similar to the finding that reaction times for solving an

Method Experiment 2

The main finding from Experiments 1A and 1B was that T2 report accuracy was impaired when T2 was included in the memory set that was retained throughout the RSVP trial. The goal of Experiment 2 was to determine whether this effect generalizes to T1. In other words, would T1 identification also be impaired when T1 is included in the memory set? Experiment 2 was identical to Experiment 1A (targets and memory set both presented in uppercase) with the exception that now T1 (and not T2) could be

Results Experiment 2

The memory probe was correctly classified on 91% of the trials. On 85% of these trials, T1 was correctly identified. As can be seen in Fig. 3a, T1 report accuracy increased across lags in all conditions. In addition, T1 performance appeared to be worst when T1 was included in the memory set, and there also appeared to be a difference between the no load and T1 not in set conditions.

A repeated measures ANOVA of T1 identification accuracy as a function of lag (1, 3, 5, or 8) and memory-load

Discussion Experiment 2

The results from Experiment 2 replicate the finding of Experiment 1 that report of RSVP targets is impaired for targets that are included in the memory set. In particular, Experiment 2 buttresses the generality of the effect observed for T2 report in Experiments 1A and 1B by showing that the same effect occurs for T1 report. In addition, the results from Experiment 2 show that even though T1 report was impaired on trials in which T1 was included in the memory set, this did not affect report of

General discussion

In three experiments, we investigated how recall of letter targets presented in RSVP is affected when the target matches a letter retained in working memory for a secondary task. In contrast to the predicted benefit for report of memorized targets, the results from all three experiments consistently showed that report accuracy was reduced for targets that were held in memory during the RSVP task. The magnitude of the impairment observed for memorized targets was stable across inter-target

Acknowledgements

The experiments reported in this study were conducted as part of Mark Nieuwenstein’s dissertation while at Utrecht University. This work was also supported by NIMH grant MH47432. We would like to thank Mary C. Potter, Sander Nieuwenhuis and an anonymous reviewer for helpful comments.

References (34)

  • D.E. Broadbent et al.

    From detection to identification: response to multiple targets in rapid serial visual presentation

    Perception and Psychophysics

    (1987)
  • F.K. Chua et al.

    Nature of the codes extracted during the attentional blink

    Journal of Experimental Psychology: Human Perception and Performance

    (2001)
  • M.M. Chun

    Types and tokens in visual processing: a double dissociation between the attentional blink and repetition blindness

    Journal of Experimental Psychology: Human Perception and Performance

    (1997)
  • M.M. Chun et al.

    A two-stage model for multiple target detection in rapid serial visual presentation

    Journal of Experimental Psychology: Human Perception and Performance

    (1995)
  • J.L. Davenport et al.

    The locus of semantic priming in RSVP target search

    Memory and Cognition

    (2005)
  • R. Desimone et al.

    Neural mechanisms of selective visual attention

    Annual Review of Neuroscience

    (1995)
  • P.E. Downing

    Interactions between visual working memory and selective attention

    Psychological Science

    (2000)
  • Cited by (0)

    View full text