Elsevier

Journal of Memory and Language

Volume 96, October 2017, Pages 122-137
Journal of Memory and Language

Working memory capacity is equally unrelated to auditory distraction by changing-state and deviant sounds

https://doi.org/10.1016/j.jml.2017.05.005Get rights and content

Highlights

  • We tested if WMC is differentially related to changing-state and deviation effect.

  • WMC was neither correlated with the deviation nor with the changing-state effect.

  • No difference was found between these correlations.

  • There is no dissociation between deviation and changing-state effect.

  • The results support a unitary over a the duplex-mechanism account of auditory distraction.

Abstract

The duplex-mechanism account states that there are two fundamentally different types of auditory distraction. The disruption by a sequence of changing auditory distractors (the changing-state effect) is attributed to the obligatory processing of the to-be-ignored information, which automatically interferes with short-term memory. The disruption by a sequence with a single deviant auditory distractor (the deviation effect), in contrast, is attributed to attentional capture. This account predicts that working memory capacity (WMC) is differentially related to the changing-state effect and to the deviation effect: The changing-state effect is assumed to be immune to cognitive control and, thus, to be unrelated to WMC. The deviation effect, in contrast, is assumed to be open to cognitive control and, thus, to be negatively related to WMC. Despite several methodological improvements over previous studies (large sample sizes, a composite measure of WMC, and a direct statistical comparison of the correlations), there was no evidence of a dissociation between the changing-state effect and the deviation effect. WMC was unrelated both to the size of the changing-state effect and to the size of the deviation effect, irrespective of whether simple stimuli (letters, Experiments 1 and 3) or complex stimuli (words and sentences, Experiment 2) were used as auditory distractors. Furthermore, a cross-experimental analysis with a total sample of N = 601 participants disconfirmed the idea that both types of auditory distraction show a differential relationship with WMC. Implications for models of auditory distraction are discussed.

Introduction

It is well established that task-irrelevant auditory stimuli disrupt working memory functions (Bell et al., 2013, Colle and Welsh, 1976, Ellermeier and Zimmer, 2014, Marsh et al., 2014, Schlittmeier et al., 2008, Tremblay and Jones, 1998). Performance is impaired although participants are required to concentrate only on the visually presented stimuli, and are instructed to ignore all incoming auditory information. Although auditory information could, in principle, be efficiently suppressed at early stages of processing in cross-modal paradigms (Guerreiro, Murphy, & Van Gerven, 2010), there is often surprisingly substantial disruption of ongoing cognitive activities. This disruption can be seen as a failure of selective attention. Individuals with problems of controlling the contents of working memory may inadvertently process information that is irrelevant for the task at hand, which may interfere with the processing of the relevant material. However, involuntary attention switching has also been described as a vital built-in mechanism that is designed to monitor the environment for signals that are potentially relevant, and to interrupt ongoing processes once such stimuli are detected. According to the latter perspective, auditory distraction is the consequence of a system that has the delicate task of balancing out the conflicting goals of focusing on task-relevant information and remaining open for information that could be of even greater importance for the individual (e.g., the sound of a fire alarm during a written exam). In the present study, we examine the relationship between working memory capacity (WMC) and two commonly examined types of auditory distraction—distraction by changing-state sounds and distraction by deviant sounds—to gain a better understanding of the nature of these effects.

The standard paradigm for examining auditory distraction is the serial recall paradigm. A key finding in this paradigm is that the immediate serial recall of visually presented targets is impaired when auditory distractors are presented during target encoding or during a short retention interval (Buchner et al., 2004, Miles et al., 1991). The amount of distraction is mainly determined by the occurrence of abrupt changes in the to-be-ignored material and not by other potentially relevant variables such as sound level (Ellermeier and Hellbrück, 1998, Ellermeier and Zimmer, 2014). Two phenomena are often distinguished. First, the changing-state effect (Bell et al., 2010, Campbell et al., 2002, Jones and Macken, 1993, Jones et al., 1992) refers to the observation that steady-state sequences consisting of repetitions of a single distractor item (e.g. A A A A A A A A) are less disruptive than changing-state sequences consisting of different distractor items (e.g. ABCDEFGH). Second, the deviation effect is caused by a violation of expectations that are based on regularities in the unfolding auditory stimulation (Hughes et al., 2007, Lange, 2005, Vachon et al., 2017). Often, the deviation effect is examined by comparing steady-state sequences to deviation sequences with a single distractor item deviating from a repetitive sequence of steady-state distractors (e.g. A A A  A B A A A).

At first glance, the changing-state effect and the deviation effect seem to be quite similar in that both effects essentially show that abrupt changes in the auditory modality disrupt serial recall. Therefore, it seems reasonable to assume that both phenomena can be attributed to the same underlying mechanism. Such a unitary explanation is offered by the embedded-processes model (Cowan, 1995), which attributes both the changing-state effect and the deviation effect to attentional capture. The model assumes that incoming stimuli are automatically compared against a neural model of the previous stimulation. If a mismatch is detected, attention is involuntarily oriented towards this mismatch. The changing-state effect can be elegantly explained by this model by assuming that changes in the auditory modality lead to some degree of attentional orienting away from the rehearsal of the target material. Obviously, the explanation of the deviation effect does not require any additional assumptions within this model.

Despite their similarities, it has been proposed that the changing-state effect and the deviation effect require fundamentally different explanations. According to the duplex-mechanism account (Hughes, 2014, Hughes et al., 2007), the changing-state effect results from an automatic conflict between the obligatory processing of the order of the discrete distractor items and the voluntary processing of the order of the target items. More precisely, it is assumed that incoming distractor sequences are automatically segmented into auditory objects when differences between adjacent distractors are detected. The order of these auditory objects is preattentively processed, and this processing interferes with the maintenance of the order of the to-be remembered material. The repetition of a single distractor item does not yield any order information and therefore does not interfere with order maintenance. The deviation effect, in contrast, is attributed to a different mechanism: attentional capture. The violation of an expectation is assumed to capture attention, which interferes with the encoding—but not with the retention—of the target items (Hughes, Vachon, & Jones, 2005).

At first glance, it might seem surprising that two phenomena that are superficially so similar do require so fundamentally different explanations. Indeed, it has been acknowledged even by proponents of the duplex-mechanism account that, “on the face of it, the unitary account is the more attractive given its obvious parsimony” (Hughes et al., 2007, p. 1052), but they argue that the acceptance of the duplex-mechanism account is necessitated by dissociations between the changing-state effect and the deviation effect that cannot be easily integrated into a unitary account (Hughes et al., 2013, Hughes et al., 2005, Hughes et al., 2007, Sörqvist, 2010; for a review see Hughes, 2014). In total, these empirical arguments are seen as so compelling that the duplex-mechanism account has become the standard model for understanding auditory distraction in recent years despite being less parsimonious than a unitary model (e.g. Elliott et al., 2016, Röer et al., 2013, Schwarz et al., 2015, Sörqvist, 2010).

Nevertheless, it has been argued that a closer look at the data reveals that the empirical basis is less compelling than often assumed (e.g. Röer et al., 2014a, Röer et al., 2015). A recurring problem is that the arguments in favor of a dissociation of the changing-state effect and the deviation effect more often than not rely on comparisons across different experimental setups that do not allow one to compare the two phenomena directly. This is not ideal for drawing conclusions because dissociations might have been produced by methodological differences between experiments rather than by differences between the changing-state effect and the deviation effect per se (see Röer et al., 2014a, for an example). These issues suggest that more direct evidence is necessary before concluding that “the distinction at the heart of the duplex-mechanism account” is necessitated by “various functional dissociations between the impact of an auditory deviation and the changing-state effect” (Hughes, 2014, p. 32).

Here, we focus on the assumption that inter-individual differences in working memory capacity (WMC) are negatively associated with the deviation effect while they are unrelated to the changing-state effect. This dissociation has been repeatedly brought forward in favor of functionally different mechanisms underlying these two effects (e.g., Hughes, 2014). The goal of the present study is to test this hypothesis, thereby overcoming some methodological problems that could have influenced the outcomes of previous studies on this issue.

Working memory is often thought to refer to a construct that provides quick access to information that is needed for ongoing cognitive processes (Wilhelm, Hildebrandt, & Oberauer, 2013). Accordingly, working memory capacity is thought to reflect inter-individual differences in the limited capacity of a person’s working memory, that is, in the amount of information individuals have available for ongoing cognitive processes. Most tasks therefore require participants to store information over a short period of time while performing other cognitive activities such as solving arithmetic problems or reading sentences (Lewandowsky et al., 2010, Oswald et al., 2014, Redick et al., 2012). For example, in a typical complex-span task such as the operation span task (Turner & Engle, 1989), participants have to evaluate the correctness of mathematical equations, each followed by the presentation of a word. After having responded to a set of these equations, the participants are prompted to recall the presented words in their correct order.

There are different theoretical views on what underlies individual differences in WMC. For example, it has been suggested that inter-individual differences in WMC largely reflect the capacity with which memory processes such as rehearsal, maintenance, updating and controlled search can be carried out (Unsworth & Engle, 2007) or, alternatively, the efficiency with which short-term memory bindings (such as the binding of an item to its list position) can be formed and maintained (Wilhelm et al., 2013). According to the executive-attention view (Engle, 2002), WMC measures the individual ability to use cognitive control to focus attention on maintaining information in working memory while avoiding distraction by concurrent cognitive activities. This theoretical view is mainly based on findings showing that WMC predicts performance in tasks that require executive control such as the Stroop task or the dichotic-listening task (Conway et al., 2001, Engle, 2002, Kane et al., 2001), and is therefore often used in the irrelevant-sound literature to justify the prediction that persons with high WMC should be less distracted by attention-grabbing sound than persons with low WMC (Elliott and Cowan, 2005, Hughes, 2014, Hughes et al., 2013, Sörqvist, 2010). However, it is sensible to note that the view that high WMC is associated with a greater ability to resist interference is not unambiguously supported by the available literature (e.g. Friedman and Miyake, 2004, Oberauer et al., 2004, Redick et al., 2011, Wilhelm et al., 2013). Furthermore, it seems obvious from the literature review presented above that the predictions of the relation between WMC and auditory distraction necessarily depend on the view of WMC that is adopted.

The embedded-processes model (Cowan, 1995) is usually interpreted as predicting that high WMC should be associated with a greater capacity to resist auditory distraction (Elliott, 2002, Elliott and Cowan, 2005, Hughes et al., 2013, Sörqvist, 2010, Sörqvist et al., 2013). Given that the model assumes that the changing-state effect and the deviation effect are both based on attentional capture, it seems reasonable to postulate that they should both be negatively correlated with WMC. This prediction is explicitly derived from the executive-attention view of WMC (Engle, 2002). However, other views of WMC lead to different predictions. Based on the view that individual differences in WMC reflect differences in mnemonic processing such as rehearsal (Unsworth & Engle, 2007), Elliott and Cowan (2005) entertained the possibility that WMC could even be positively related to the amount of distraction by irrelevant auditory distractors. Given that it is likely that auditory distraction interferes with the rehearsal of the target items (Röer, Bell, & Buchner, 2014b), distraction effects may be more pronounced in individuals with high WMC who show more rehearsal of the target items than individuals with low WMC because more rehearsal may provide more opportunity for disruption (Elliott & Cowan, 2005).

While it is not as easy as it may appear at first to derive a clear prediction from the embedded-processes model (Cowan, 1995) without making further assumptions about the nature of WMC, the model clearly does not predict that there should be any differences in the relationship between WMC and the changing-state effect on the one hand and WMC and the deviation effect on the other hand because both effects are attributed to the same process (attentional capture). Therefore, individual differences in WMC should be similarly related to both types of auditory distraction.

The duplex-mechanism account (Hughes, 2014, Hughes et al., 2013, Sörqvist, 2010), in contrast, predicts a differential relationship between WMC and the changing-state effect on the one hand and WMC and the deviation effect on the other hand. The key assumption of this account is that there is “a distinction between two forms of auditory distraction—one controllable by the individual, the other less so, if at all” (Hughes, 2014, p. 37f). One of the defining differences between the changing-state effect and the deviation effect is that the latter should be negatively related to WMC while the former should be unrelated to WMC (Hughes et al., 2013, Sörqvist, 2010). The changing-state effect is postulated to be immune to cognitive control because the effect is assumed to be underpinned by automatic, obligatory processing of order information that is not accessible to cognitive control, and does not involve attentional mechanisms. The processing of the order of the to-be-ignored distractors should be obligatory in individuals with high and low WMC. This leads to the prediction that the changing-state effect must be unrelated to WMC. Due to the greater involvement of attention, the deviation effect, in contrast, should be more open to top-down cognitive control than the changing-state effect (Parmentier & Hebrero, 2013). Individuals with high WMC should be better at voluntarily suppressing the bottom-up orientation of attention towards the deviant distractors than individuals with low WMC who should be more easily distracted (Hughes et al., 2013, Sörqvist, 2010). This leads to the prediction that the changing-state effect and the deviation effect should be differentially related to WMC. This is one of the dissociations that form the empirical basis of the duplex-mechanism account (Hughes, 2014).

In the past, the literature has often been presented as supporting the idea of a differential relationship of WMC to the changing-state effect and the deviation effect (Hughes, 2014). However, while it has been demonstrated repeatedly that distraction by changing-state sequences (consisting of letters, words, non-words, or tones) does not correlate with WMC (Beaman, 2004, Parmentier and Hebrero, 2013, Sörqvist, 2010, Sörqvist et al., 2013), the hypothesis of a negative relationship between the deviation effect and WMC can be challenged for several reasons. Due to the grater involvement of attention, the deviation effect, in contrast, should be more open to top-down cognitive control than the changing-state effect. One reason is that previous studies yielded inconsistent results. While three experiments (Hughes et al., 2013, Sörqvist, 2010) showed that the deviation effect (operationalized as the difference between the steady-state and the deviation condition) was negatively correlated to performance in the operation span task, a later study with a primary focus on age differences in distraction failed to replicate this result: In this study, WMC was found to correlate neither with the changing-state effect nor with the deviation effect (Röer, Bell, Marsh, & Buchner, 2015).

Another reason is that there are methodological issues as well that seem to necessitate further research. (1) The studies providing supporting evidence for a relationship between the deviation effect and WMC had only small to medium sample sizes (N = 24 in the experiment examining the deviation effect and N = 31 in the experiment examining the changing-state effect in the study of Hughes et al., 2013; N = 40 in Experiment 1 and N = 48 in Experiment 2 of Sörqvist, 2010), whereas the study that showed a null correlation had a sample size that was almost twice as large (N = 258) as the combined sample size of the other four studies. This is important because sample correlations are known to be variable and inaccurate in small samples and gradually stabilize at the level of the population correlation as the sample sizes increase (Schönbrodt & Perugini, 2013). Furthermore, studies with small samples often provide exaggerated estimates of effect sizes (Button et al., 2013). To avoid these problems, comparatively large samples were used in the present study. Obviously, a larger sample size should increase the statistical power to find a relationship between WMC and auditory distraction if it existed. (2) In all previous studies only a single WMC measure was reported: performance in the operation span task. Specific WMC tasks such as the operation span task may measure task-specific variance that is unrelated to the construct of interest (Conway et al., 2005, Lewandowsky et al., 2010). For instance, the operation span task may reflect not only WMC, but also arithmetic capability. Therefore, it is often recommended to combine multiple measures (e.g., operation span and sentence span) into a composite WMC score to obtain a more general estimate of WMC with better psychometric properties (Conway et al., 2005, Wilhelm et al., 2013). In the present study, we used two different complex span tasks from a well-validated standardized working memory test battery with good psychometric properties (Lewandowsky et al., 2010). This approach should have further increased our chances to find a relationship between WMC and auditory distraction if it existed. (3) Importantly, the hypothesis of a differential relationship of changing-state effect and deviation effect to WMC was not directly tested in previous studies although the main prediction of the duplex-mechanism account is that “there are fundamental differences between the changing-state effect and aspecific attentional capture” (Hughes, 2014, p. 33). In the study of Hughes et al. (2013), changing-state effect and deviation effect were examined in different experiments, which makes a direct comparison of the correlations difficult. In the study of Sörqvist (2010), the conclusion that “the relationship between WMC and the deviation effect is significantly different from the relationship between WMC and the changing-state effect“ (p. 657) was based on the finding that WMC correlated significantly with the deviation effect while the corresponding correlation between WMC and the changing-state effect did not attain significance. This interpretation is problematic because “the difference between ‘significant’ and ‘not significant’ is not itself statistically significant” (Gelman & Stern, 2006, p. 328). To illustrate, a data pattern where one correlation just reaches the statistical significance threshold (e.g., with p = 0.04) while the other just falls short of significance (e.g., with p = 0.06) does not provide conclusive evidence of a statistically significant dissociation. The relevant statistical test is whether the correlation between WMC and the deviation effect is significantly different from the correlation between WMC and the changing-state effect. This test was not reported. Without the relevant statistical test, it is difficult to draw clear conclusions from these studies (Diedenhofen and Musch, 2015, Nieuwenhuis et al., 2011). In fact, the difference between the correlations observed by Hughes et al. (2013) would not have reached significance even when tested with a one-sided test, z = 1.55, p = 0.06.1 This means that there is no clear “psychometric evidence for the dissociation between the two forms of auditory distraction” (Hughes et al., 2013, p. 549), but a p-value of 0.06 in combination with small sample sizes also does not provide clear evidence against it either. Given that the available evidence does not allow us to draw clear conclusions about this issue, there is need for further research in which this hypothesis is tested directly.

In the present study, we applied a statistical test of significance between correlations (between WMC and the changing-state effect on the one hand and WMC and the deviation effect on the other hand). This allowed us to directly test the central prediction of the duplex-mechanism account that there should be a dissociation between the changing-state effect and the deviation effect. More precisely, the duplex-mechanism account predicts that there is a negative relationship between WMC and the deviation effect while the relationship between WMC and changing-state distraction is absent. This leads to the statistical prediction that the correlation between WMC and the deviation effect should be significantly different from the correlation between WMC and the changing-state effect.

Section snippets

Participants

A total of 138 students at Heinrich Heine University Düsseldorf (95 women) with a mean age of 24 years (SD = 4.88) participated in exchange for course credit or a small honorarium. All participants were fluent German speakers and reported normal hearing and normal or corrected-to-normal vision.

Working memory tasks

In order to minimize the task-specific variance associated with single complex span tasks (Conway et al., 2005, Lewandowsky et al., 2010), we applied two different complex span tasks: the operation span task

Participants

Sixty-three students at Heinrich Heine University Düsseldorf (42 women) with a mean age of 25 years (SD = 5.99) participated in exchange for course credit or a small honorarium. All were fluent German speakers and reported normal hearing and normal or corrected-to-normal vision.

Materials and procedure

Materials and procedure were identical to those of Experiment 1 with the following exceptions. The changing-state sequences consisted of 24 German sentences (as in the studies of Bell, Röer, Dentale, & Buchner, 2012; and of

Participants

One hundred and forty-two students at Heinrich Heine University Düsseldorf (110 women) with a mean age of 22 years (SD = 3.69) participated in exchange for course credit or a small honorarium. All were fluent German speakers and reported normal hearing and normal or corrected-to-normal vision.

Materials and procedure

Given that Experiment 3 was a close replication of Experiment 1 of Sörqvist (2010), only the key aspects of this experiment are described here. All participants started with the working memory task, followed

General discussion

One of the most widely accepted accounts of auditory distraction to date—the duplex-mechanism account (Hughes, 2014)—is based on the assumption that there are two functionally distinct types of auditory distraction. The changing-state effect is assumed to be due to automatic processing of the distractor material, and, therefore, to be unaffected by cognitive control. The deviation effect is assumed to be caused by attentional orienting, which is assumed to be open to cognitive control. One

Acknowledgements

The research reported in this article was funded by the Deutsche Forschungsgemeinschaft (BE 4311/3-1).

References (72)

  • R. Bell et al.

    Habituation of the irrelevant sound effect: Evidence for an attentional theory of short-term memory disruption

    Journal of Experimental Psychology: Learning Memory and Cognition

    (2012)
  • A. Buchner et al.

    Valence of distractor words increases the effects of irrelevant speech on serial recall

    Memory & Cognition

    (2004)
  • K.S. Button et al.

    Power failure: Why small sample size undermines the reliability of neuroscience

    Nature Reviews Neuroscience

    (2013)
  • T. Campbell et al.

    Auditory memory and the irrelevant sound effect: Further evidence for changing-state disruption

    Memory

    (2002)
  • A.R.A. Conway et al.

    The cocktail party phenomenon revisited: The importance of working memory capacity

    Psychonomic Bulletin & Review

    (2001)
  • A.R.A. Conway et al.

    Working memory span tasks: A methodological review and user's guide

    Psychonomic Bulletin & Review

    (2005)
  • N. Cowan

    Attention and memory: An integrated framework

    (1995)
  • L.J. Cronbach et al.

    How we should measure “change“: Or should we?

    Psychological Bulletin

    (1970)
  • B. Diedenhofen et al.

    Cocor: A comprehensive solution for the statistical comparison of correlations

    PLoS ONE

    (2015)
  • W. Ellermeier et al.

    Is level irrelevant in “irrelevant speech“? Effects of loudness, signal-to-noise ratio, and binaural unmasking

    Journal of Experimental Psychology: Human Perception and Performance

    (1998)
  • W. Ellermeier et al.

    Individual differences in susceptibility to the “irrelevant speech effect”

    Journal of the Acoustical Society of America

    (1997)
  • W. Ellermeier et al.

    The psychoacoustics of the irrelevant sound effect

    Acoustical Science and Technology

    (2014)
  • E.M. Elliott

    The irrelevant-speech effect and children: Theoretical implications of developmental change

    Memory & Cognition

    (2002)
  • E.M. Elliott et al.

    Individual differences in the ability to avoid distracting sounds

    European Journal of Cognitive Psychology

    (2006)
  • E.M. Elliott et al.

    Coherence of the irrelevant-sound effect: Individual profiles of short-term memory and susceptibility to task-irrelevant materials

    Memory & Cognition

    (2005)
  • R.W. Engle

    Working memory capacity as executive attention

    Current Directions in Psychological Science

    (2002)
  • F. Faul et al.

    G∗Power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences

    Behavior Research Methods

    (2007)
  • N.P. Friedman et al.

    The relations among inhibition and interference control functions: A latent-variable analysis

    Journal of Experimental Psychology: General

    (2004)
  • A. Gelman et al.

    The difference between “significant” and “not significant” is not itself statistically significant

    The American Statistician

    (2006)
  • L.D. Goodwin et al.

    Understanding correlation: Factors that affect the size of r

    Journal of Experimental Education

    (2006)
  • M.J.S. Guerreiro et al.

    The role of sensory modality in age-related distraction: A critical review and a renewed view

    Psychological Bulletin

    (2010)
  • J.B. Hittner et al.

    A Monte Carlo evaluation of tests for comparing dependent correlations

    The Journal of General Psychology

    (2003)
  • R.W. Hughes

    Auditory distraction: A duplex-mechanism account

    PsyCh Journal

    (2014)
  • R.W. Hughes et al.

    Cognitive control of auditory distraction: Impact of task difficulty, foreknowledge, and working memory capacity supports duplex-mechanism account

    Journal of Experimental Psychology: Human Perception and Performance

    (2013)
  • R.W. Hughes et al.

    Auditory attentional capture during serial recall: Violations at encoding of an algorithm-based neural model?

    Journal of Experimental Psychology: Learning Memory and Cognition

    (2005)
  • R.W. Hughes et al.

    Disruption of short-term memory by changing and deviant sounds: Support for a duplex-mechanism account of auditory distraction

    Journal of Experimental Psychology: Learning, Memory, and Cognition

    (2007)
  • Cited by (35)

    View all citing articles on Scopus
    View full text