Elsevier

Neuropsychologia

Volume 48, Issue 7, June 2010, Pages 2167-2173
Neuropsychologia

Electrophysiological correlates of proactive interference in the ‘Recent Probes’ verbal working memory task

https://doi.org/10.1016/j.neuropsychologia.2010.04.008Get rights and content

Abstract

Using event-related potentials (ERPs), the present study examined the temporal dynamics of proactive interference in working memory using a recent probes task. Participants memorized and retained a target set of four letters over a short retention interval. They then responded to a recognition probe by judging whether it was from the memory set. ERP waveforms elicited by positive probes compared to those from negative probes showed positive shifts in a fronto-central early N2 component and a parietal late positive component (LPC). The LPC was identified as the electrophysiological signature of proactive interference, as it differentiated between two types of negative probes defined based on whether they were recently encountered. These results indicate that the proactive interference we observed arises from a mismatch between familiarity and contextual information during recognition memory. When considered together with related studies in the literature, the results also suggest that there are different forms of proactive interference associated with different neural correlates.

Introduction

Much modern research in cognitive psychology and cognitive neuroscience has been devoted to working memory, which is considered to be the crucial interface between long-term knowledge representations and perception (Baddeley and Hitch, 1974, Jonides et al., 2008). A prominent feature of working memory is that only information pertinent to present task demands should be maintained and used but not irrelevant or outdated information, even though they may have been relevant moments before (Bjork, 1978, Roth and Courtney, 2007). Consequently, an important research topic is to understand how proactive interference (PI) arises when information that was once maintained in working memory interferes with current processing and how such interference is resolved.

Whereas classical PI research relied primarily on measures of recall accuracy, Monsell (1978) introduced a ‘Recent Probes’ paradigm based on the Sternberg item-recognition task (1966), allowing the examination of PI with response times. In this paradigm, participants hold a small set of target items in memory over a short retention interval and then make a yes (positive) or no (negative) decision to a recognition probe. Critically, a probe can be drawn from the target set of the immediately preceding trials, making it a ‘Recent probe’. In contrast, a ‘Non-recent probe’ has not been encountered recently. Accurate no decisions have been found to be significantly slower to Recent Negative probes than to Non-recent Negative probes, indicating a PI effect that memory for items recently encountered interferes with current item-recognition (Monsell, 1978). Starting from the work of Jonides, Smith, Marshuetz, and Koeppe (1998), researchers have adopted this paradigm in a number of brain imaging studies to identify brain structures associated with PI resolution in working memory, such as left inferior frontal gyrus (Badre and Wagner, 2005, Bunge et al., 2001, D’Esposito et al., 1999, Mecklinger et al., 2003, Yi et al., 2009).

To complement these PET and fMRI localization studies for a full elucidation of the mechanisms of proactive interference, more research is needed to reveal the corresponding temporal dynamics. In the present study, we applied event-related potentials (ERPs), a cognitive neuroscience technique that excels in temporal resolution to the recent probes verbal working memory task. Specifically, we compared ERP responses elicited by Recent Negative probes with those elicited by Non-recent Negative probes to identify ERPs that differ between these two conditions, namely, the electrophysiological signature of PI. Previously, there were only three related ERP studies (Du et al., 2008, Tays et al., 2008, Tays et al., 2009).

The Du et al. (2008) study adopted a working memory paradigm involving directed forgetting that Zhang, Leung, and Johnson (2003) introduced. In this paradigm, participants held in memory an initial set of letters and were then cued to ignore some of the letters and to remember the rest as the final target set for probe recognition. As in the recent probes tasks, the critical comparison between Recent Negative probes drawn from the ignored subset and Non-recent Negative probes drawn from letters not shown in the present trial has revealed reliable behavioral PI effects in several studies adopting this paradigm (Nee et al., 2007, Yi et al., 2009, Zhang et al., 2003).

Using ERP, Du et al. (2008) observed a fronto-central negative component peaking around 300 ms post-probe-onset whose amplitude was significantly smaller for the Recent Negative probes than for the Non-recent Negative ones. They interpreted this finding as an N2 effect indexing the resolution of proactive interference induced by familiarity. This interpretation was reasonable in the sense that their N2 resembles an FN400 component that is often thought to index familiarity in recognition memory (Curran, 2000, Mecklinger, 2000). It becomes questionable, however, in light of more recent evidence indicating that FN400 may be associated not with familiarity but rather with conceptual priming (for reviews, see Paller et al., 2007, Voss and Paller, 2008), or processes downstream from those responsible for computing familiarity information (Tsivilis, Otten, & Rugg, 2001). Regardless of its interpretation, the observation in Du et al. that N2 is sensitive to recency manipulation suggests a possible prediction for the present study. Assuming that the PI effects found in different research paradigms are associated with the same neural signature, one would predict the same ERP component to differ across the two Negative conditions in the present recent probes task involving a similar recency manipulation. That is, a smaller N2 would be expected for the Recent Negative condition compared to the Non-recent Negative condition.

Alternatively, one could predict PI effects to be associated with later ERP components, in light of a biased-competition model of proactive interference as discussed in Jonides and Nee (2006). The model was initially proposed by Desimone and Duncan (1995) to explain the mechanism of visual selective attention and later developed by Kan and Thompson-Schill (2004) to also explain conceptual selection. By this model, proactive interference is resolved in the same evaluation process that distinguishes between positive and negative probe items, and the decision of whether a probe belongs to the current target set is based on competition between two types of memory codes, familiarity and contextual information. For the recent probes task, the Recent Negative probes, which have a higher level of familiarity due to their recent appearance in the preceding trials, are more likely to prompt a yes response than Non-recent Negative probes. To overcome this inappropriate response tendency, the decision-making process must be controlled and biased to rely more on the correct contextual information.

According to this model, the recency manipulation in the present study should modulate ERP components associated with a processing stage in which both familiarity and contextual information are available and can thus interact with each other for the proposed competition. The late positive component (LPC) seems to be such a candidate, as it is known to index both familiarity and episodic retrieval in long-term memory recognition (Voss and Paller, 2008, Voss and Paller, 2009). There are reasons to assume that this is true in short-term memory as well. First, at a general level, short- and long-term memories have been argued to be in fact the same system (for review, see Jonides et al., 2008). Second, there are both theoretical models and empirical evidence suggesting that familiarity and episodic recollection as dual processes for long-term recognition (e.g., Yonelinas, 1994) are also at work in short-term memory recognition (Crites et al., 2000, Crites et al., 1998, Danker et al., 2008, McElree and Dosher, 1989, Oberauer, 2008, Öztekin and McElree, 2007). Therefore, if it is assumed that LPC in short-term memory is like its counterpart in long-term memory in indexing both familiarity and episodic retrieval of contextual information, the Jonides and Nee model (2006) would predict the LPC to be modulated by recency manipulation in the present study, reflecting the competition between familiarity and contextual information.

Thus, we designed an experiment to induce proactive interference with a recent probes task and to identify the ERPs associated with the interference effect. Based on the literature, the analysis emphasized the early N2 component and the late positive component.

Section snippets

Participants

Nineteen college students (10 males, 9 females, age range from 20 to 26 years, mean ± SD = 22.9 ± 1.9 years) participated in this study. All were right-handed with normal or corrected-to-normal vision, and none reported any history of neurological or psychiatric disorders. Written informed consent was obtained from each participant in accordance with guidelines of the Institute of Psychology, Chinese Academy of Sciences, Beijing. Data from one participant were discarded due to excessive eye blinks

Behavioral data

Although our design distinguished between Recent Positive and Non-recent Positive conditions as in Jonides et al. (1998), the first-pass analysis did not reveal any significant differences between these two trial types for either the behavioral data (mean RT: 676 ms vs. 668 ms, mean error rate: 11.1% vs. 13.4%) or the EEG data. For simplicity, the two positive conditions were merged in the following analysis.

Table 1 shows the behavioral results for all three conditions. An ANOVA of the response

Discussion

The behavioral results showed significantly faster responses in the positive condition than in the negative condition1

Conclusion

The present study examined the neural correlates of PI in working memory using ERPs. ERP waveforms associated with memory recognition were found to distinguish between positive and negative probes in an early N2 component and the LPC. The LPC but not the N2 component differentiated between two types of negative probes defined based on whether they were recently encountered, and was therefore identified as the electrophysiological signature of proactive interference. When considered together

Acknowledgments

This research was supported by the National Natural Science Foundation of China (#30670702) and a grant to Young Scientists from Institute of Psychology, Chinese Academy of Sciences (#O9CX032003). We thank two anonymous reviewers for their insightful and valuable comments on earlier versions of the paper.

References (45)

  • W.J. Tays et al.

    General proactive interference and the N450 response

    Neuroscience Letters

    (2009)
  • D. Tsivilis et al.

    Context effects on the neural correlates of recognition memory: An electrophysiological study

    Neuron

    (2001)
  • J.L. Voss et al.

    Brain substrates of implicit and explicit memory: The importance of concurrently acquired neural signals of both memory types

    Neuropsychologia

    (2008)
  • R. West et al.

    Event-related neural activity associated with the stroop task

    Cognitive Brain Research

    (1999)
  • C.C. Woodruff et al.

    Electrophysiological dissociation of the neural correlates of recollection and familiarity

    Brain Research

    (2006)
  • J.X. Zhang et al.

    Frontal activations associated with accessing and evaluating information in working memory: An fMRI study

    NeuroImage

    (2003)
  • A. Baddeley et al.

    Working memory

  • D. Badre et al.

    Frontal lobe mechanisms that resolve proactive interference

    Cerebral Cortex

    (2005)
  • S.A. Bunge et al.

    Prefrontal regions involved in keeping information in and out of mind

    Brain

    (2001)
  • S.L. Crites et al.

    Immediate and delayed stimulus repetitions evoke different ERPs in a serial-probe recognition task

    Psychophysiology

    (2000)
  • S.L. Crites et al.

    Event related potentials and serial position effects in a visual probe recognition task

    Psychophysiology

    (1998)
  • T. Curran

    Brain potentials of recollection and familiarity

    Memory & Cognition

    (2000)
  • Cited by (12)

    • Electrophysiological signatures of attentional control in bilingual processing: Evidence from proactive interference

      2021, Brain and Language
      Citation Excerpt :

      As a result, positive and negative deflections are obtained and its amplitude, latency and topography reflect underlying cognitive processes related to the specific conditions and stimuli of the task. Previous studies have recognized a negative ongoing ERP component (N2) at around 300 ms (Du et al., 2008) and a late positive component (LPC) at around 700 ms, as the ERP signature of PI using verbal stimuli (Zhang, Wu, Kong, Weng, & Du, 2010). N2 has been interpreted as an index of various processes in cognitive control such as cue monitoring, conflict resolution, inhibition and attentional control (Folstein & Van Petten, 2008).

    • Mechanisms and neuronal networks involved in reactive and proactive cognitive control of interference in working memory

      2014, Neuroscience and Biobehavioral Reviews
      Citation Excerpt :

      The differences between the studies of Du et al. (2008) and Zhang et al. (2010) may be interpreted at least partially as a consequence of the different manipulations to induce interference in working memory (intra-trial interference in a directed forgetting manipulation versus recent negative probes). In contrast, the studies of Tays and co-workers (2008) as well as Zhang and co-workers (2010) both used the recent negative task. Nonetheless, Tays et al. (2008) favored the familiarity-inhibition mechanism, whereas Zhang et al. (2010) proposed the late acting context retrieval mechanism.

    • Electrophysiological indices of interference resolution covary with individual fluid intelligence: Investigating reactive control processes in a 3-back working memory task

      2014, NeuroImage
      Citation Excerpt :

      However, the scalp location and the patterns of activation were consistent, leading to the conclusion that the MFN reflected the recruitment of cognitive control mechanisms. These conclusions were further supported by source localization procedures which pointed to the anterior cingulate (Liotti et al., 2000) and prefrontal cortices (Tays et al., 2009, 2008; West and Alain, 1999, 2004; Zhang et al., 2010) as generators of the scalp MFN. Nonetheless, the EEG studies mentioned above have two main limitations: first, they do not take into account possible differences in electrocortical activity among individuals.

    View all citing articles on Scopus

    This study was carried out in Institute of Psychology, Chinese Academy of Sciences, Beijing, China.

    View full text