Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2021 Jan 15.
Published in final edited form as: Audit Percept Cogn. 2020 Nov 10;3(1-2):18–32. doi: 10.1080/25742442.2020.1842996

Decomposing the role of rehearsal in auditory distraction during serial recall

Angela M AuBuchon 1, Corey I McGill 2, Emily M Elliott 2
PMCID: PMC7810201  NIHMSID: NIHMS1643584  PMID: 33458602

Abstract

According to the interference-by-process mechanism of auditory distraction, irrelevant changing sounds interfere with subvocal articulatory-motor sequencing during rehearsal. However, previous attempts to limit rehearsal with concurrent articulation and examine the residual irrelevant sound effect have limited both cumulative rehearsal as well as the initial assembly of articulatory-phonological labels. The current research decomposed rehearsal into these two levels of articulatory-phonological sequencing: silent concurrent articulation limits the availability of both serial repetition and articulatory-phonological recoding; rapid serial visual presentation allows for articulatory-phonological recoding but presents items too quickly for cumulative serial repetition. As predicted by the interference-by-process account, concurrent articulation -- but not rapid serial visual presentation -- reduced the irrelevant sound effect. Not only did the irrelevant sound effect persist in the face of rapid serial visual presentation, a steady-state effect also emerged. These findings indicate that irrelevant sounds interfere with both serial processing at the level of articulatory-motor planning at the word level as well as in the formation of item-to-item associations created via serial repetition of complete items. Moreover, these findings highlight the benefits of articulatory-phonological recoding – independent of pure rehearsal -- within serial recall.

Keywords: Irrelevant Sound Effect, Rehearsal, Attention, Concurrent Articulation, Articulatory Suppression

Rehearsal, the silent repetition of to-be-remembered stimuli, is a useful strategy for maintaining verbal materials during short-term memory serial recall tasks (Atkinson & Shiffrin, 1968; Baddeley, Thomson, & Buchanan, 1975). However, imposing seemingly simple distractors, such as a processing task between items or irrelevant sounds, appears to disrupt rehearsal processes (e.g., Cowan, 2001; Jones & Macken, 1993). During a conventional irrelevant sound effect (ISE) task, participants are given instructions to ignore the auditory stimuli (e.g. spoken letters “T - B - K - R - M”) while trying to remember a visually-presented list of items (e.g. digits); moreover, information from the auditory modality is never called upon to complete the task. Because the irrelevant sounds disrupt performance despite their explicit lack of relevance to the focal memory task, the interference-by-process account proposes that the auditory stimuli cause interference by gaining obligatory access to the same systems which process the visually-presented memoranda (e.g. Hughes, Vachon, & Jones, 2007; Hughes, Chamberland, Tremblay, & Jones, 2016).

In the case of serial recall of visually-presented material, irrelevant auditory stimuli are assumed to interfere within the serial processing system that carries out rehearsal, (Jones & Tremblay, 2000; but see Norris, Baddeley, & Page, 2004). However, Souza and Oberauer (2018) recently observed that adults’ serial recall performance did not benefit from instructions to rehearse in a cumulative fashion intended to promote serial order retention (e.g. Item 1, Item 1, Item 2, Item 1, Item 2, Item 3, etc). If rehearsal benefits serial retention less than originally thought, then it is unclear why disrupting rehearsal with irrelevant sounds would impair performance. This raises the question of whether irrelevant sounds truly disrupt rehearsal or, instead, disrupt another serial order process related to rehearsal. In order to identify when in the rehearsal process irrelevant sounds first cause disruption, the current research decomposed rehearsal processes into two subcomponents, (1) an initial recoding of visual information into an articulatory or phonological code as well as (2) the cumulative, serial repetition of that code.

Serial Order and Irrelevant Sounds

A fundamental characteristic of auditory input is the temporal ordering of acoustic features. The interference-by-process mechanism of auditory distraction specifies that auditory input gains pre-attentive, obligatory access to the same serial processing system recruited by rehearsal (Elliott, Hughes, Briganti, Joseph, Marsh, & Macken, 2016; Hughes, 2014; Hughes et al., 2007; Jones & Macken, 1993). Evidence that auditory distraction disrupts rehearsal often comes from manipulating the serial processing demands of the memory tasks. Conventional ISE tasks require correct serial recall of the visually-presented lists. However, an adaptation which required recall of only a single item per list demonstrated that auditory distraction is lessened when maintenance of serial order is not required (Beaman & Jones, 1997; Elliott et al., 2016; Klapp, Marshburn, & Lester, 1983). In the single-item recall adaptations, lists were presented as in a conventional ISE paradigm: participants saw lists of nine digits (from the set 0–9) in the presence of auditory distractors. However, two types of recall trials were tested: during “probed-order” recall, participants were probed with a single digit cue from the preceding list and asked to provide the digit that followed the cue. During “missing-item” recall, participants were asked to report which digit (0 thru 9) had not appeared in the preceding nine-item list. Thus, participants needed to maintain both item and order information to successfully complete a probed-order trial. In contrast, participants needed only item information to successfully complete a missing-item trial. Adults displayed a much smaller auditory distraction effect during the missing-item trials than during the probed-recall trials in the presence of changing irrelevant sound, a finding consistent with interference to rehearsal (Beaman & Jones, 1997; Elliott et al., 2016). Moreover, participants who reported spontaneous engagement of rehearsal during the missing-item task continued to be disrupted by irrelevant sound -- despite not needing to maintain serial order; in contrast, participants who reported not engaging rehearsal showed no such changing state effect (Hughes & Marsh, 2020).

The use of encoding and retrieval strategies that deemphasize retention of serial order also reduces the ISE when the entire list must be recalled. Even when serial recall is not required, participants often still use rehearsal as a maintenance strategy and recall items in the same order they were presented. Thus, when whole list recall is required – even when serial order is not – an ISE is observed (Beaman & Jones, 1998; Perham, Banbury & Jones, 2007). However, when participants are required to organize list recall based on semantic categories of the items – thus, deemphasizing serial recall and the usefulness of a rehearsal strategy -- irrelevant sounds no longer produced an observable disruption to memory performance (Perham et al., 2007).

The findings from missing item tasks as well as free and categorical recall, are taken as evidence that using rehearsal to maintain serial order is a key determinant in whether or not an ISE will emerge. However, strong evidence for order interference does not equate to strong evidence for rehearsal interference, in part because rehearsal is traditionally observed only indirectly (see Lewandowsky & Oberauer, 2015). This inability to directly measure rehearsal limits the understanding of the relative importance of rehearsal as a serial order mechanism disrupted in the ISE. For example, rather than directly manipulating access to rehearsal, Hughes and Marsh (2020) separated rehearsers from non-rehearsers post hoc based on self-reported strategy use. Although no method is currently available to directly measure rehearsal, certain manipulations are thought to limit rehearsal use during immediate recall tasks.

Preventing Rehearsal with Concurrent Articulation

The exact nature of rehearsal is still under debate. Many models of short-term and working memory describe rehearsal as a form of silent speech which benefits memory via serial maintenance of decaying verbal-phonological codes (Atkinson & Shiffrin, 1968; Baddeley et al., 1975) -- with some models emphasizing rehearsal’s role in reactivating the phonological trace within attention (Cowan, 2001). In contrast, the interference-by-process account specifies that rehearsal benefits memory, not by serial reactivation, but by creating item-to-item associations during repetitive subvocal articulatory-motor sequencing (Hughes et al, 2016).

Despite their disagreements on the nature of the form being rehearsed (i.e. a verbal-phonological trace or an articulatory-motor plan), there is agreement that requiring participants to engage in concurrent articulation is the most “direct” method of limiting rehearsal (Bhatarah, Ward, Smith, Haynes, 2009; Larsen & Baddeley, 2003). Concurrent articulation is the overt repetition of an unrelated word or syllable during serial recall. Concurrent articulation is thought to occupy either the phonological store (e.g. Baddeley et al., 1975) or the articulatory pathways (e.g., Awh et al, 1996) that could otherwise be used for rehearsal.

Limiting rehearsal with concurrent articulation substantially reduces overall serial recall performance; additionally, it diminishes both the word length effect and the phonological similarity effect -- two well-replicated phenomena thought to arise from the use of verbal labels (Baddeley et al., 1975; Coltheart & Langdon, 1998). There is evidence that concurrent articulation also eliminates the ISE, further supporting the claim that auditory distractors interfere with serial order rehearsal processes (Hanley, 1997; Klatte, Lee, & Hellbrück, 2002; Salamé & Baddeley, 1982). However, an unresolved limitation surrounding concurrent articulation prevents these data from fully supporting an interpretation of the irrelevant sound effect in which changing auditory distractors specifically interfere with the underlying serial nature of rehearsal.

Regardless of whether one views rehearsal as a cyclical continuation of the articulatory-motor plan or as serial reactivation of phonological traces, visual memoranda must be transformed before rehearsal can be engaged. Concurrent articulation prevents both this initial transformation as well as any subsequent serial rehearsal (Hanley, 1997). The ISE is often observed when stimuli are presented aurally and the need to recode is minimal (though see Bosen, Monzingo & AuBuchon, 2020). However, unlike with visual presentation in which concurrent articulation significantly reduces the ISE, concurrent articulation only minimally reduces the size of the ISE when stimuli are presented aurally. This finding suggests that articulatory-phonological recoding is independently impacted by both concurrent articulation and irrelevant sound (see Hanley, 1997 for a review). Notably, speech production requires a serial processor for the coordination of articulatory-motor sequences, even at the word level. Therefore, according to the interference-by-process account, irrelevant sounds may compete with the peripheral articulatory motor-sequencing processes required for recoding even when cumulative rehearsal is not in use.

The Current Study

The goal of the current experiment was to isolate the processes of recoding from the process of cumulative rehearsal as one possible basis of order interference from irrelevant sound. Prior research has established that irrelevant sound disrupts maintenance for serial order and has assumed rehearsal as the target mechanism. Direct evidence that serial recall in short-term memory is susceptible to irrelevant sound even when speech-motor and/or phonological sequencing is in use only during recoding would be strong evidence in support of the interference-by-process mechanism. The current study aims to isolate the initial assembly of an articulatory-phonological code from the cyclical, serial repetition of that code which would signify rehearsal.

The process of labeling aloud a visually presented digit takes adults, on average 333 ms (Wagner, Torgesen, Rashotte, & Pearson, 1999). Presenting visual stimuli in rapid succession should allow participants time to transform each item into either an articulatory and/or phonological assembly while limiting their ability to complete any serial sub-vocal recitations before the presentation of the next item (Bunting, Cowan, & Saults, 2006). Therefore, we manipulated the availability of cumulative rehearsal during an ISE task through two rehearsal-limiting methodologies, namely silent concurrent articulation (which would prevent both recoding and rehearsal) and speeded presentation (which would allow for recoding, but minimize any serial repetition).

Predictions

We anticipate that, consistent with previous findings, silent concurrent articulation will reduce the overall ISE1. This outcome would support the interpretation that irrelevant sound gains obligatory access to a serial processor otherwise used by rehearsal. However, it would be unclear whether the point of interference was at articulatory-phonological encoding or serial repetition. Rapid serial visual presentation should minimize serial repetition but allow recoding. Because recoding relies on articulatory-motor and/or phonological sequencing, the re-emergence of the ISE during rapid visual presentation would support an interpretation in which even a minimal amount of sequencing can be disrupted by irrelevant sound.

Exploratory investigation of the role of attention in the ISE

Interference-by-process is one of two mechanisms described by the duplex-mechanism account. Attentional diversion also appears to play a role in auditory distraction during serial recall (Hughes, 2014). To explore possible independent contributions of serial processes and attention, we included two types of irrelevant sounds: steady-state and changing-state sounds. Repeating the same auditory token at a fixed pace (i.e. steady-state) introduces minimal variability into the auditory stream so places low demands on serial processing. Therefore, the effect of steady-state sounds (compared to quiet) could be used to index the contribution of attentional diversion processes to the disruptive effects of irrelevant sound (Bell, Röer, Lang, & Buchner, 2019). Although irrelevant steady-state sounds are capable of producing significant effects on children’s serial order recall (AuBuchon, McGill, & Elliott, 2018), steady-state sounds typically result in small effects on adults’ serial order recall. Such small effects are often difficult to detect (e.g. Bell et al., 2019; Elliott & Briganti, 2012; Jones et al., 1993; Lange, 2005) further highlighting the importance of serial processing in the irrelevant sound effect (e.g. Jones et al., 1993). For adults, disruption to performance typically occurs when the auditory distractors noticeably change from one token to the next (i.e. changing-state effect). Thus, the overall ISE can be deconstructed into the steady-state effect (SSE) indexing attention and the changing-state effect (CSE) indexing disruption to serial processing described in the interference-by-process account.

Because the attentional and interference-by-process components can be manipulated independently of one another within the serial recall paradigm by separately examining their respective components of the ISE (e.g., Hughes et al., 2013), secondary predictions can be made regarding changing- and steady-state effects: based on previous research with adults, we predict that under conventional ISE testing procedures, the steady-state effect (SSE; performance in silence minus performance in steady-state sounds) will be small and indistinguishable from zero with the current sample size (e.g., Bell et al., 2019). Thus, the entirety of the disruption observed in conventional trials will be due to changing-state sounds, and we will examine this by calculating both the ISE described above as well as the changing-state effect (CSE; performance in steady-state sounds minus performance in changing-state sounds). If either manipulation is successful at isolating the common reliance on a serial processor, then changes to the size of the auditory disruption effects under the rehearsal-limiting conditions should be reflected only in this changing-state component. Specifically, the expected CSE should be observed when rehearsal is available (i.e. under conventional testing procedures); however, the CSE should be reduced to zero when the affected component of rehearsal is prevented. Because both manipulations are intended to affect rehearsal, in neither case would the interference-by-process account predict a change in the SSE.

Methods

Participants

132 undergraduate psychology students (86 female; 36 male; 10 unknown/not reported) at Louisiana State University aged 18 to 27 participated for course credit. All participants reported normal or correct-to-normal vision, normal hearing, and being a native speaker of English. Five of the 132 participants did not complete the study or failed to provide responses on the memory trials. One participant reported not following the silent articulation instructions. Therefore, the analysis included 126 participants.

Procedure

All participants provided informed consent prior to the experiment in accordance with protocols approved by the local IRB. The experiment lasted approximately 25 minutes. Responses were typed using the numeric keypad. Participants completed the tasks in groups of 2–4, each at their own workstation. The workstations were divided by sound-attenuating panels that created open cubicles and were equipped with Sennheiser headphones connected to Dell Desktop computers running E-prime 2.0 experimental presentation software (Psychology Software Tools, Pittsburgh, PA). As part of the instructions, individuals in the silent concurrent articulation condition were encouraged to honestly answer a manipulation check to determine if they followed the articulation instruction on each relevant trial. An experimenter was present in the room during each experimental session and monitored the participants for compliance.

ISE Task

In all trials, participants were to remember a sequence of eight visually presented digits. Consistent with a conventional ISE task, each list of digits was accompanied by either silence, steady-state sounds, or changing-state sounds. Half of the trials additionally included a rehearsal-limiting manipulation: some participants engaged in silent concurrent articulation during list presentation; for other participants, the list was presented at four times the conventional speed. Lists were scored as the proportion of digits recalled in the correct serial position.

Participants completed six practice trials (one of each possible condition combination) before beginning the 60 critical trials. Critical trials were grouped into 20 blocks of three trials, one trial from each sound condition: silence, steady-state sounds, or changing-state sounds. The 10 conventional and 10 rehearsal-limiting blocks were mixed, and participants were informed of the condition before each block.

During silent trials, no additional sounds were played during the immediate serial recall task. During steady-state trials, a male voice saying the letter “Q” was played at the onset of every visually presented digit. During changing-state trials, a different randomly selected letter (“A”, “C”, “G”, “I”, “L”, “O”, “Q”, “R”, “U”) from the same male speaker was played at the onset of each visually presented digit. The sounds were presented at a comfortable listening level that was judged to be subjectively equal across each of the workstations. The experimenters checked the volume levels of each station prior to the beginning of each session and maintained a constant volume level throughout the duration of the experiment.

Participants’ rehearsal was limited in one of two ways: concurrent articulation (N = 46) or rapid serial visual presentation (N = 80). Individuals in the concurrent articulation condition were instructed to silently recite the alphabet from “A” to “Z” during the presentation of the to-be-remembered digits. Overt concurrent articulation has been previously shown to eliminate the ISE (Hanley, 1997; Salamé & Baddeley, 1982). There is some debate as to whether this disruption is due to a de facto ISE created by the concurrent articulation adding irrelevant sounds to the “silent” condition of the paradigm or simply because of the additional demands of vocalizing (Gupta & MacWhinney, 1995; Neath, Farley, and Surprenant, 2003). Therefore, we required silent concurrent articulation as it occupies the same articulatory motor planning pathways as overt articulation without adding irrelevant sounds to the auditory environment or requiring overt vocalizations (Gupta & MacWhinney, 1995). As a manipulation check, participants in the silent concurrent articulation group were asked after each trial if they completed the silent articulation component as requested. For the individuals in the speeded presentation condition, the to-be-remembered items in the rehearsal-limiting trials were presented every 250ms instead of every 1000ms, with no blank screen between successive digits.

A 1000 ms presentation rate is common in studies of the ISE. At this presentation rate, adults are observed spontaneously using cumulative rehearsal strategies at early serial positions on about 25% of trials, and rehearse subsets of serially adjacent items (e.g. 2–3, 3–4) towards the end of the list on over 50% of trials (Souza & Oberauer, 2018). A 250 ms presentation rate is comparable to the 84%ile on the Rapid Digit Naming subtest on the Comprehensive Test of Phonological Processing in which adults are to name aloud a list of digits as quickly as possible (Wagner et al., 1999). Therefore, the 250 ms rate was intended to limit most adults’ ability to rehearse the to-be-remembered information while still allowing participants time to silently transform the presented item into an articulatory-phonological code. Notably, despite the rehearsal limiting presentations, both tasks required maintenance of serial order and neither introduced additional acoustic features into the auditory environment.

Analysis Plan

On the silent concurrent articulation manipulation check, one participant reported adhering to the silent concurrent articulation instructions on only 4 of 30 trials, so was excluded. The participants included in the analysis indicated failure to silently articulate on an average of 0.41 (range of 0 to 5) trials across all auditory conditions. Those trials were also excluded from the analysis. The remaining data were analyzed using Bayesian ANOVAs and T-tests from the BayesFactor package (version 0.9.12–4.2; Morey & Rouder, 2018) in R (RStudio Team, 2018)2. Bayesian ANOVA provides Bayes factors favoring all possible combinations of model parameters relative to the null (or “no difference”) model. Unless specified, all t-tests were directional with the alternative model representing an effect greater than zero versus a point null. The Bayes factor quantifies the strength of evidence in favor of one model relative to the other. Bayes factors larger than 3 indicate support for the alternative model, Bayes factors between 1 and 3 are considered inconclusive, and Bayes factors less than 1 indicate support for the null model. Thus, Bayes factors provide a useful alternative to traditional significance testing when anticipated results include support for the null (or “no difference”) model (Rouder, Morey, Verhagen, Swagman, & Wagenmakers, 2017). In the present study, we anticipate that there should be no CSE when rehearsal is limited by concurrent articulation or rapid serial visual presentation. Such a claim cannot be supported in traditional significance testing. Each model was estimated with 100,000 samples. Otherwise, the default prior settings were used.

We entered the proportion correct into Bayesian ANOVAs with the group (silent concurrent articulation or rapid serial visual presentation) as a between-participants factor and rehearsal presence (conventional or limiting) as well as irrelevant sound type (silence, steady-state, changing-state) as within-participants factors. We followed this omnibus analysis with two separate analyses: First, we tested a model of only the conventional trials to ensure that we replicated the patterns of auditory distraction reported in previous literature. Then, we directly tested our prediction that the CSE would decrease to zero by calculating the difference scores analogous to the CSE (i.e., performance during steady sounds – performance during changing sounds) in both the conventional and rehearsal limiting conditions. The estimate of the CSE was then entered into a directional Bayesian T-test against the point null.

Results

To assess the overall pattern of results, proportion correct scores were entered into a Bayesian ANOVA with group as a between-participants factor and rehearsal presence and irrelevant sound type as within-participants factors. The best ANOVA model included main effects of rehearsal presence and irrelevant sound type, as well as the 2-way interaction of group and rehearsal presence, BF = 1.74 × 1061, ± 0.80% -- though a more complex model which additionally included a 3-way interaction of group, rehearsal presence, and irrelevant sound had similar levels of support, BF = 1.49 × 1061, ± 1.24% (Figure 1).

Figure 1:

Figure 1:

Open in a new tab

Mean proportion correct for each auditory condition under conventional presentation (black bars) and under rehearsal-limiting constraints (grey bars). The group instructed to use silent concurrent articulation as their rehearsal-limiting constraint is on the left (N=46). The group presented with rapid serial visual presentation as their rehearsal-limiting constraint is on the right (N=80). Error bars represent standard error of the mean.

Replication of Conventional Auditory Distraction Effects.

Before testing our predictions, it was crucial to confirm that both groups displayed the expected auditory distraction effect. Therefore, a second Bayesian ANOVA was conducted only on the conventional trials. Group was a between-participants factor and irrelevant sound type was a within-participants factor. The best ANOVA model included only a main effect of irrelevant sound type, BF = 1.06 × 108, ± 0.66%. This model was preferred more than 2:1 over the model which additionally included a main effect of group and preferred 11:1 over the model with an interaction of group and irrelevant sound type. Thus, the evidence indicated that the two groups exhibited similar auditory distraction effects in conventional trials (Figure 1). In order to understand the main effect of irrelevant sound, we compared models including one of three ways of coding that factor: (1) the three-level coding distinguishing between silence, steady sounds, and changing sounds, (2) coding irrelevant sound such that silence was contrasted with steady and changing sounds, and (3) coding irrelevant sound such that silence and steady sound were contrasted with changing sound. Evidence favored the third coding structure over the first coding structure by a factor of more than 3 and the second coding structure by a factor of more than 400,000. Thus, as is typically observed in adult studies of the irrelevant sound effect, memory for visually-presented materials did not noticeably decrease in the presence of irrelevant steady-state sounds but did decrease in the presence of irrelevant changing-state sounds.

Bayesian t-tests were used to examine individual ISEs and their components (i.e. the Steady-State Effect and the Changing State effect). Consistent with the ANOVA reported above, in the conventional trials the evidence supported ISEs greater than zero when neither concurrent articulation (BF > 5000; M=0.082; SD=0.108) nor rapid serial visual presentation (BF > 2000; 0.059; SD=0.114) were present. Based on prior literature on adults, we predicted that the ISE was driven by the changing-state sounds with minimal disruption due to steady-state sounds. Accordingly, CSEs were greater than zero when neither concurrent articulation (BF = 141; 0.059; SD=0.104) nor rapid serial visual presentation (BF = 125; M=0.052; SD=0.125) were present. In contrast, the directional Bayesian T-tests examining the SSEs favor the nulls for both the concurrent articulation (BF = 0.69; M=0.023; SD=0.111) and rapid serial visual presentation (BF = 0.21; M=0.007; SD=0.106) groups.

Auditory Distraction Effects under Silent Concurrent Articulation.

The primary goal of this experiment was to determine if limiting rehearsal via concurrent articulation and/or rapid serial visual presentation would also eliminate--or at least reduce--the ISE observed in the conventional trials. Moreover, the duplex-mechanism account specifies that interference-by-process is indexed by the CSE. Therefore, any declines in the overall ISE when rehearsal is limited should be mirrored in the CSE. Consistent with this prediction, when rehearsal was limited by concurrent articulation, evidence favored the point null model (BF = 0.32) over the directional alternative that the ISE was greater than 0. This finding was replicated in the CSE: evidence favored the point null model, BF = 0.24, indicating that the CSE decreased to zero in the presence of silent concurrent articulation. However, it is worth noting that despite both the mean ISE (M=0.0126; SD=0.112) and mean CSE (M=0.009; SD=0.124) being indistinguishable from 0 in the presence of silent concurrent articulation, individual scores covered similar ranges under rehearsal limitations (ISE: −0.26 to 0.31; CSE −0.29 to 0.35) and during conventional testing (ISE: −0.23 to 0.39; CSE: −0.18 to 0.25; see Supplementary Figure 1).

Auditory Distraction Effects under Rapid Serial Visual Presentation.

In contrast to the findings with concurrent articulation, when rehearsal was limited with rapid serial visual presentation, the ISE was still greater than zero (BF > 1 billion). In an exploratory analysis, we created two alternative models: one in which the ISE under RSVP (M=0.090, SD=0.105) was greater than zero but less than 0.059 -- the group’s mean ISE on the conventional trials -- and one in which the ISE under RSVP was greater than 0.59. The model in which the ISE was greater than 0.059 was preferred over the model in which the ISE was less than 0.059 by a factor of more than 3 billion. Thus, not only did the ISE re-emerge when articulatory motor planning was available, but the ISE also appeared to increase. Separate examinations of the CSE and SSE suggested that this increase was due to both an increase in the CSE as well as the emergence of a SSE. When rehearsal was limited with rapid serial visual presentation, the CSE was still greater than zero, BF > 73,000. Using the same exploratory procedure described above, the model in which the CSE under RSVP (M=0.061; SD=0.098) was greater than under conventional testing (i.e. M=0.052; SD=0.124) was preferred by a factor of more than 400 million. Additionally, evidence from the directional Bayesian T-tests supported an SSE (M=0.029; SD=0.094) greater than zero (BF = 8.82).

Discussion

The current experiment utilized two different methods of limiting rehearsal within a standard irrelevant sound paradigm. The goal was to decompose rehearsal into an initial articulatory-phonological recoding stage and the serial repetition of that code – with the latter considered “rehearsal” in the classical sense. Not only does this approach allow us to more narrowly identify that the recoding stage of sequential processing is disrupted by irrelevant sound, but it also provides insight into how verbal processes benefit serial recall. Concurrent articulation is a well replicated method of limiting rehearsal, but this manipulation would also prevent in the initial recoding of visual stimuli into an articulatory-phonological code. In contrast, speeded presentation allows for the initial recoding stage but limits any subsequent repetition of the new code.

Consistent with previous findings, when rehearsal was limited with concurrent articulation -- albeit without overtly speaking -- participants displayed, on average, no negative effects of irrelevant sounds (Gupta & MacWhinney, 1995). This finding is consistent with the interference-by-process mechanism in which both rehearsal and the obligatory processing of changing irrelevant sounds are thought to rely on similar serial order processing. However, even in the silent condition, concurrent articulation reduced recall to .40 (see Figure 1). This low level of recall means that participants recalled, on average, 3.2 items from the 8-item list. Given that 3 items is the lower bound of the capacity of attention (Cowan, 2001), it is unsurprising that a manipulation such as irrelevant sound did not further disrupt performance in either changing- or steady-state sounds (see also, Neath, 2000). This observation highlights the need for new approaches to study the relationship between rehearsal and other cognitive processes which affect memory.

Despite the compelling findings at the group level, participants displayed wide variability, covering the range of −0.26 to 0.38 (Supplemental Figure 1). This was surprising as we might have assumed that limiting rehearsal would minimize individual differences by forcing all participants to use a similar mnemonic strategy. For example, when rehearsal is limited by overt current articulation, adaptive forward and backward serial recall became moderately correlated suggesting that participants used similar strategies on both tasks (AuBuchon, Stone, Kronenberger, & Pisoni, 2020). However, these correlations only emerged when the stimuli were pictures of concrete objects or novel symbolic shapes; forward and backward recall of visually presented digits remained only weakly correlated. The authors concluded that, because digits are highly familiar and strongly associated with one another, some participants could automate phonological recoding and rehearsal strategies during forward span despite being engaged in concurrent articulation. However, requiring both concurrent articulation and list reversal appeared to sufficiently limit these verbal strategies.

A similar interpretation may be made of the current results. Extensive rehearsal use might lead to rapid automatization of the rehearsal process -- especially with digits -- making those participants who efficiently engage rehearsal strategies less vulnerable to disruption by either concurrent articulation or changing irrelevant sounds. In contrast, inefficient rehearsal may yield few order cues leaving inefficient rehearsers susceptible to disruption by either. Finally, participants who do not engage rehearsal at all should show no disruption from either concurrent articulation or changing irrelevant sounds. Such a hypothesis would predict a curvilinear relationship with neither nonrehearsers nor highly efficient rehearsers showing changing-state effects, and maximal changing-state effects displayed by people attempting to rehearse but doing so inefficiently3. This possible interpretation would explain some of the individual differences observed in the conventional trials -- at least the ISE values greater than zero. How this proposed curvilinear relationship might be affected in the presence of both concurrent articulation and irrelevant sound is less clear. In fact, 35% of participants displayed negative effects (better performance in the presence of noise than in silence) when engaged in silent concurrent articulation. These findings highlight the need to better understand both individual differences in strategy selection as well as individual differences in the efficiency in which the selected strategy is implemented.

Of particular interest was that rapid serial visual presentation did not reduce the size of the ISE. Rapidly presenting digits provided insufficient time for effective cumulative rehearsal; therefore, if serial repetition was the sole component of rehearsal impacted by auditory distraction, then the ISE would have been reduced as was observed with silent concurrent articulation. However, the presentation rate did provide sufficient time for articulatory-phonological recoding. Therefore, both assembling an articulatory/phonological sequence and the cyclical, serial repetition of that sequence are individually impacted by irrelevant sounds. Specifically, both irrelevant sound and articulatory-phonological sequencing compete for a shared serial process.

An unexpected finding was that the steady-state effect (SSE) emerged during rapid serial visual presentation. It is not obvious how an interference-by-process account would explain this particular aspect of the results. AuBuchon et al. (2018) interpreted the SSE as a marker of attentional capture in children. If that interpretation is extended here, it is likely that rapid serial visual presentation requires more attentional resources than the standard rate of presentation in adults. This interpretation contradicts earlier work claiming that increasing task difficulty promotes task engagement and leaves individuals less susceptible to distraction from auditory deviants (Hughes, Hurlstone, March, Vachon & Jones, 2013). Pilot data collected by Hughes et al. (2013) confirmed that partially obscuring the visually-presented memoranda slowed encoding speed and increased task difficulty; however, presentation rate during their memory task exceeded the encoding speed estimates they observed in the pilot data, and accuracy was not impacted in either the encoding speed pilot task or the serial recall task. In contrast, rapid presentation during the current study decreased overall performance, even in silence (Figure 1), suggesting that the changes in task difficulty are not comparable between the two studies. Therefore, the influence of task difficulty on susceptibility to auditory distraction should investigated further, along with the individual differences considerations that were discussed above. Minimally, this finding highlights the contextual and individual differences which influence attentional disruption due to irrelevant sound and reinforces the need to include measures of both attentional capture and interference-by-processes to better understand potential dependencies.

Clearly, additional research is needed to resolve discrepancies in the existing literature, and we advocate strongly that all three auditory conditions be included in future designs (i.e., silence, steady- and changing-state sounds). For example, Kattner and Ellermeier (2019) demonstrated a significant decrease in the size of the ISE in a sample of adults who had undergone five days of training in a dichotic listening task, while the active control group did not show any decrease in the size of the ISE. This decrease in the ISE illustrates that experience controlling attention mediates the size of the ISE and is important evidence in favor of attentional accounts of auditory distraction. However, the change in performance in the proportional measure of irrelevant speech compared to white noise was from 16.4% to 10.7%. Thus, while there was a decrease, the ISE was not eliminated, suggesting other processes beyond attentional diversion were contributing to the effect. Also, Kattner and Ellermeier did not include a steady state condition which means that the SSE and the CSE could not be independently evaluated. Nonetheless, the combination of the current finding of a significant SSE with rapid presentation, the findings of AuBuchon et al. (2018), and the training study of Kattner and Ellermeier (2019) all point to a role for both articulatory-phonological sequencing and attention-based processes within the ISE, and provide important evidence for boundary conditions of when these processes are most likely to exert their influence.

In conclusion, rehearsal has long been implicated as the serial process disrupted by changing-state sounds. However, prior experiments using concurrent articulation to limit rehearsal during visual presentation have limited both the cyclical serial repetition of articulatory-phonological labels as well as the initial assembly of those labels. The current experiment teased apart these two stages of labeling and rehearsal. The findings indicate that the serial process disrupted in the presence of irrelevant sound can occur at the level of articulatory-phonological sequencing. Moreover, the finding highlights the benefits to serial recall by simply labeling the visually-presented items. Additionally, an unexpected emergence of the SSE under rapid serial visual presentation as well as the large range of ISEs and CSEs under concurrent articulation raise further questions about the influences of contextual and individual differences during auditory distraction.

Supplementary Material

Supplemental Material

Supplemental Figure 1: Histograms of the ISE (right) and CSE (left) for the silent concurrent articulation group: (a) ISE under conventional testing, (b) CSE under conventional testing, (c) ISE under silent concurrent articulation, (d) CSE under silent concurrent articulation.

ACKNOWLEDGEMENTS

The authors would like to thank Rebecca Wagner for assistance with data management.

FUNDING DETAILS

AA’s contribution to this research was supported, in part, by the National Institute of General Medical Sciences of the National Institutes of Health under award number P20 GM109023.

Footnotes

1

For the purposes of the current paper, the term “ISE” will be used to refer specifically to the comparison of serial recall performance in the presence of a changing-state sound, relative to serial recall performance in silence; however, it is important to note that the irrelevant sound stimuli may vary across studies (e.g., such as using a non-speech sound like a tone changing pitch; Jones & Macken, 1993).

2

Corresponding frequentist tests were consistent with the Bayesian ones. Frequentist models returned significant p-values for factors and interactions included in winning models in the Bayesian analysis and null p-values when the null model was favored.

3

We would like to thank an anonymous reviewer for critical input on developing this argument.

DECLARATION OF INTEREST STATEMENT

The authors have no conflicts of interest to disclose. These data were also presented at Psychonomic Society’s 57th Annual Meeting in Boston, Massachusetts, November, 2016.

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are openly available in Open Science Framework (OSF) at https://osf.io/fdqbv.

REFERENCES

  1. Atkinson RC, & Shiffrin RM (1968). Human memory; A proposed system and its control processes. Psychology of Learning and Motivation, 2, 89–195. doi: 10.1016/S0079-7421(08)60422-3 [DOI] [Google Scholar]
  2. AuBuchon AM, Kronenberger WG, Stone L, & Pisoni DB (2020). Strategy Use on Clinical Administrations of Short-Term and Working Memory Tasks. Journal of Psychoeducational Assessment, 0734282920930924.
  3. AuBuchon AM, McGill CI, Elliott EM (2018). Auditory distraction does more than disrupt rehearsal processes in children’s serial recall. Memory and Cognition, 47(4), 738–748. 10.3758/s13421-018-0879-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Awh E, Jonides J, Smith EE, Schumacher EH, Koeppe RA, & Katz S (1996). Dissociation of Storage and Rehearsal in Verbal Working Memory: Evidence from Positron Emission Tomography. Psychological Science, 7(1), 25–31. 10.1111/j.1467-9280.1996.tb00662.x [DOI] [Google Scholar]
  5. Baddeley AD, Thomson N, & Buchanan M (1975). Word length and the structure of short-term memory. Journal of Verbal Learning and Verbal Behavior, 14, 575–589. 10.1016/S0022-5371(75)80045-4 [DOI] [Google Scholar]
  6. Beaman C, & Jones DM (1997). Role of Serial Order in the Irrelevant Speech Effect: Tests of the Changing-State Hypothesis. Journal of Experimental Psychology: Learning, Memory, and Cognition, 23(2), 459–471. [Google Scholar]
  7. Beaman CP, & Jones DM (1998). Irrelevant sound disrupts order information in free recall as in serial recall. The Quarterly Journal of Experimental Psychology: Section A, 51(3), 615–636. [DOI] [PubMed] [Google Scholar]
  8. Bell R, Röer JP, Lang A-G, Buchner A (2019). Distraction by steady-state sounds: Evidence for a graded attentional model of auditory distraction. Journal of Experimental Psychology: Human Perception and Performance, 45, 500–512. 10.1037/xhp0000623 [DOI] [PubMed] [Google Scholar]
  9. Bhatarah P, Ward G, Smith J, & Hayes L (2009). Examining the relationship between free recall and immediate serial recall: Similar patterns of rehearsal and similar effects of word length, presentation rate, and articulatory suppression. Memory & Cognition, 37(5), 689–713. doi: 10.3758/MC.37.5.689. [DOI] [PubMed] [Google Scholar]
  10. Bosen AK, Monzingo E, & AuBuchon AM (2020). Acoustic-phonetic mismatches impair serial recall of degraded words. Auditory Perception & Cognition, 1–21. 10.1080/25742442.2020.1846012 [DOI] [PMC free article] [PubMed]
  11. Bunting M, Cowan N, & Saults JS (2006). How does running memory span work?. Quarterly journal of experimental psychology (2006), 59(10), 1691–1700. doi: 10.1080/17470210600848402 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Coltheart V, & Langdon R (1998). Recall of short word lists presented visually at fast rates: Effects of phonological similarity and word length. Memory & Cognition, 26, 330–342. [DOI] [PubMed] [Google Scholar]
  13. Cowan N (2001). The magical number 4 in short-term memory: A reconsideration of mental storage capacity. Behavioral and Brain Sciences, 24, 87–185 [DOI] [PubMed] [Google Scholar]
  14. Elliott EM, & Briganti AM (2012). Investigating the role of attentional resources in the irrelevant speech effect. Acta Psychologica, 140(1), 64–74. doi: 10.1016/j.actpsy.2012.02.009. [DOI] [PubMed] [Google Scholar]
  15. Elliott EM, Hughes RW, Briganti A, Joseph TN, Marsh JE, & Macken B (2016). Distraction in verbal short-term memory: Insights from developmental differences. Journal of Memory and Language, 88, 39–50. doi: 10.1016/j.jml.2015.12.008. [DOI] [Google Scholar]
  16. Gupta P, MacWhinney B (1995). Is the Articulatory Loop Articulatory or Auditory? Reexamining the Effects of Concurrent Articulation on Immediate Serial Recall. Journal of Memory and Language, 34, 63–88. [Google Scholar]
  17. Hanley JR (1997). Does articulatory suppression remove the irrelevant speech effect? Memory, 5, 423–431. doi: 10.1080/741941394 [DOI] [PubMed] [Google Scholar]
  18. Hughes RW (2014). Auditory distraction: A duplex mechanism account. PsyCh Journal, 3(1), 30–41. doi: 10.1002/pchj.44. [DOI] [PubMed] [Google Scholar]
  19. Hughes RW, Chamberland C, Tremblay S, & Jones DM (2016). Perceptual-motor determinants of auditory-verbal serial short-term memory. Journal of Memory and Language, 90, 126–146. [Google Scholar]
  20. Hughes RW, Hurlstone MJ, Marsh JE, Vachon F, & Jones DM (2013). 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, 39(2), 539–553. doi: 10.1037/a0029064. [DOI] [PubMed] [Google Scholar]
  21. Hughes RW, & Marsh JE (2020). When is forewarned forearmed? Predicting auditory distraction in short-term memory. Journal of Experimental Psychology: Learning, Memory, and Cognition, 46(3), 427–442. doi: 10.1037/xlm0000736. [DOI] [PubMed] [Google Scholar]
  22. Hughes RW, Vachon F, & Jones DM (2007). 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, 33(6), 1050–1061. doi: 10.1037/0278-7393.33.6.1050. [DOI] [PubMed] [Google Scholar]
  23. Jones DM, & Macken WJ (1993). Irrelevant tones produce an irrelevant speech effect: Implications for phonological coding in working memory. Journal of Experimental Psychology: Learning, Memory, and Cognition, 19(2), 369–381. doi: 10.1037/0278-7393.19.2.369. [DOI] [Google Scholar]
  24. Jones DM, Macken WJ, & Murray AC (1993). Disruption of visual short-term memory by changing-state auditory stimuli: The role of segmentation. Memory & Cognition, 21(3), 318–328. doi: 10.3758/BF03208264. [DOI] [PubMed] [Google Scholar]
  25. Jones DM, & Tremblay S (2000). Interference in memory by process or content? A reply to Neath (2000). Psychonomic Bulletin & Review, 7, 550–558. doi: 10.3758/BF03214370 [DOI] [PubMed] [Google Scholar]
  26. Kattner F, & Ellermeier W (2019). Distraction at the cocktail party: Attenuation of the irrelevant speech effect after a training of auditory selective attention. Journal of Experimental Psychology: Human Perception and Performance. doi: 10.1037/xhp0000695. [DOI] [PubMed]
  27. Klapp ST, Marshburn EA, & Lester PT (1983). Short-term memory does not involve the “working memory” of information processing: The demise of a common assumption. Journal of Experimental Psychology: General, 112, 240–264. [Google Scholar]
  28. Klatte M, Lee N, & Hellbrück J (2002). Effects of irrelevant speech and articulatory suppression on serial recall of heard and read materials. Psychologische Beiträge, Band, 44, 166–186. [Google Scholar]
  29. Lange EB (2005). Disruption of attention by irrelevant stimuli in serial recall. Journal of Memory and Language, 53(4), 513–531. doi: 10.1016/j.jml.2005.07.002. [DOI] [Google Scholar]
  30. Larsen JD, & Baddeley A (2003). Disruption of verbal STM by irrelevant speech, articulatory suppression, and manual tapping: Do they have a common source? The Quarterly Journal of Experimental Psychology: Human Experimental Psychology, 56A(8), 1249–1268. doi: 10.1080/02724980244000765. [DOI] [PubMed] [Google Scholar]
  31. Lewandowsky S, & Oberauer K (2015). Rehearsal in serial recall: An unworkable solution to the nonexistent problem of decay. Psychological Review, 122(4), 674. [DOI] [PubMed] [Google Scholar]
  32. Morey RD, & Rouder JN (2018). BayesFactor: Computation of Bayes Factors for Common Designs. R. package version 0.9.12–4.2. https://CRAN.R-project.org/package=BayesFactor
  33. Rouder JN, Morey RD, Verhagen J, Swagman AR, & Wagenmakers EJ (2017). Bayesian analysis of factorial designs. Psychological Methods, 22(2), 304. [DOI] [PubMed] [Google Scholar]
  34. Neath I (2000). Modeling the effects of irrelevant speech on memory. Psychonomic Bulletin & Review, 7, 403–423. doi: 10.3758/bf03214356 [DOI] [PubMed] [Google Scholar]
  35. Neath I, Farley LA, & Surprenant AM (2003). Directly assessing the relationship between irrelevant speech and articulatory suppression. Quarterly Journal of Experimental Psychology: Section A, 56(8), 1269. doi: 10.1080/02724980244000756 [DOI] [PubMed] [Google Scholar]
  36. Norris DG, Baddeley AD & Page MPA (2004). Retroactive effects of irrelevant speech on serial recall from short-term memory. Journal of Experimental Psychology: Learning, Memory, and Cognition, 30(5), 1093. [DOI] [PubMed] [Google Scholar]
  37. Perham N, Banbury SP, & Jones DM (2007). Reduction in auditory distraction by retrieval strategy. Memory, 15(4), 465–473. [DOI] [PubMed] [Google Scholar]
  38. Psychology Software Tools, Inc. [E-Prime 2.0]. (2007). Pittsburgh, PA [Google Scholar]
  39. RStudio Team (2018). RStudio: Integrated Development for R. RStudio, Inc, Boston, MA: URL http://www.rstudio.com/. [Google Scholar]
  40. Rouder JN, Morey RD, Verhagen J, Swagman AR, & Wagenmakers E-J (2017). Bayesian analysis of factorial designs. Psychological Methods, 22(2), 304–321. 10.1037/met0000057 [DOI] [PubMed] [Google Scholar]
  41. Salamé P, & Baddeley AD (1982). Disruption of short-term memory by unattended speech: Implications for the structure of working memory. Journal of Verbal Learning and Verbal Behavior, 21(2), 150–164. 10.1016/S0022-5371(82)90521-7 [DOI] [Google Scholar]
  42. Souza AS, & Oberauer K (2018). Does articulatory rehearsal help immediate serial recall?. Cognitive psychology, 107, 1–21. [DOI] [PubMed] [Google Scholar]
  43. Wagner RK, Torgesen JK, Rashotte CA, & Pearson NA (1999). Comprehensive test of phonological processing: CTOPP. Austin, TX: Pro-ed. [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplemental Material

Supplemental Figure 1: Histograms of the ISE (right) and CSE (left) for the silent concurrent articulation group: (a) ISE under conventional testing, (b) CSE under conventional testing, (c) ISE under silent concurrent articulation, (d) CSE under silent concurrent articulation.

NIHMS1643584-supplement-Supplemental_Material.zip (116.6KB, zip)

Data Availability Statement

The data that support the findings of this study are openly available in Open Science Framework (OSF) at https://osf.io/fdqbv.

RESOURCES