Exploring the role of attention during implicit memory retrieval
Introduction
Memory retrieval is a frequent occurrence during the course of a typical day. Given that daily demands dictate that we often juggle several tasks simultaneously, it is likely that some memory retrieval occurs whilst doing something else at the same time (e.g., maintaining a conversation with someone while simultaneously trying to remember who exactly they are!). A common belief is that people cannot perform two tasks concurrently as well as they could perform each task independently. With this in mind, it is important to understand the effects of distraction on remembering.
In laboratory studies, the effects of distraction are assessed by comparing a divided attention (DA) condition, in which participants carry out a memory task while simultaneously performing a secondary task, and a full attention (FA) condition, in which participants perform only the memory task. Attention can be divided during the encoding phase, the retrieval phase, or both. It has long been recognized that dividing attention during memory encoding has a large, deleterious effect on subsequent performance (see Mulligan (2008), for review). Examining attention during retrieval, however, is a more recent development in the literature and has yielded mixed results with some studies finding large DA effects (e.g., Fernandes and Moscovitch, 2000, Fernandes and Moscovitch, 2003, Hicks and Marsh, 2000, Lozito and Mulligan, 2006) and others little effect (e.g., Baddeley et al., 1984, Craik et al., 1996).
Initial research found that DA had little effect on memory retrieval prompting the conclusion that memory retrieval is largely automatic (e.g., Baddeley et al., 1984). Subsequent research by Craik, Naveh-Benjamin and colleagues disputed this by showing that although the secondary task failed to affect retrieval, retrieval produced large and consistent costs to secondary task performance (Craik et al., 1996, Naveh-Benjamin et al., 2000). The secondary task costs imply that retrieval requires attentional resources, prompting Craik et al. (1996) to conclude that retrieval is attention-demanding yet obligatory (Anderson et al., 1998, Craik et al., 2000, Craik et al., 1996, Naveh-Benjamin and Guez, 2000, Troyer and Craik, 2000). In contrast, other studies have found large, deleterious effects of DA on recall and recognition (Carrier and Pashler, 1995, Fernandes and Moscovitch, 2000, Fernandes and Moscovitch, 2003, Hicks and Marsh, 2000, Jacoby, 1991, Lozito and Mulligan, 2006), showing that at least under some circumstances, memory retrieval can be disrupted by DA.
Fernandes and Moscovitch, 2000, Fernandes et al., 2005 documented one such set of conditions. These researchers varied the match between the materials used on the memory test (words) and the materials used in the secondary task (words or digits). These authors found that when the materials used on the memory test and secondary task matched (i.e., both used word stimuli), DA significantly reduced recall and recognition, but when the materials were of different types (e.g., a word-based memory test and a number-based secondary task), memory performance was not reduced (a pattern referred to as material-specific interference). Other work suggests that the type of retrieval is critical (Hicks and Marsh, 2000, Lozito and Mulligan, 2006), with retrieval primarily driven by recollection being highly susceptible to DA, and familiarity-driven retrieval less so.
Although a growing body of research has explored the role of attention during memory retrieval, a critical limitation is that virtually all of this research has focused on explicit memory. Little research has focused on the role of attention during implicit retrieval. Explicit memory refers to conscious or intentional recollection of prior events, assessed with tests such as recall and recognition; implicit memory refers to unintentional or unconscious influences of memory, typically assessed on tests of repetition priming (Mulligan, 2003b, Schacter, 1987).
Implicit retrieval is dissociable into perceptual and conceptual forms. Perceptual priming reflects implicit memory for stimulus form and includes tests such as word-stem completion and perceptual identification. Conceptual priming reflects implicit memory for stimulus meaning and includes the test of category-exemplar production. Because perceptual tests reflect visual or auditory information about a stimulus, such tests are general sensitive to changes between study and test items in perceptual information, such as changes in modality (visual vs. auditory) or form (e.g., picture vs. word), and generally insensitive to conceptual encoding manipulations. Because conceptual priming is heavily reliant on semantic information about the items, it generally exhibits the opposite pattern. Conceptual tests are sensitive to encoding manipulations that emphasize semantic features of stimuli and are insensitive to perceptual manipulations such as changes in study modality (see Mulligan, 2003b, Roediger and McDermott, 1993, for reviews). Importantly, dividing attention at study has dissociated perceptual and conceptual implicit tests. DA at study consistently reduces conceptual priming (Gabrieli et al., 1999, Mulligan, 1997, Mulligan, 1998, Mulligan and Hartman, 1996) whereas perceptual priming is less likely to be disrupted by the same DA manipulations (Mulligan, 2003a, Mulligan and Hartman, 1996, Mulligan and Hornstein, 2000, Russo and Parkin, 1993).
Another distinction, orthogonal to the perceptual-conceptual distinction, differentiates between production and identification priming (Fleischman et al., 2001, Gabrieli et al., 1999, LaVoie and Faulkner, 2008). Identification priming tasks instruct participants to identify a test stimulus, or verify an attribute of the stimulus. Identification tests can involve the analysis of either the form or the meaning of a stimulus, which may be presented normally or in degraded form. An example is perceptual identification, in which participants attempt to identify a fragmented or briefly presented stimulus. In identification tests, the retrieval cue and task requirements determine a single correct response. Production priming tasks are those in which test cues do not uniquely define the information to be retrieved but merely delimit a class of possible correct answers. An example is the word-stem completion task in which the word stems (e.g., sto__) match many legitimate completions (e.g., store, stone, stole, storm, etc.). For production tests, cues map onto (or activate) many potential responses.
The identification—production distinction is supported by a number of dissociations (e.g., Fleischman et al., 2001, Gabrieli et al., 1999, LaVoie and Faulkner, 2008). Critically for present purposes, it has been argued that identification and production tasks differ in attentional demands during retrieval (Fleischman et al., 2001). According to this view, production tests require the selection of a response from an array of potential responses, eliciting a degree of response competition not present in identification tests. Increased response competition is claimed to induce heavier attentional demands during retrieval for production tests relative to identification tasks (Fleischman et al., 2001, Gabrieli et al., 1999). Gabrieli et al. (1999) demonstrated that dividing attention during study affected production tasks more than identification tasks (see also, Mulligan & Peterson, 2008). However, the notion that production tests require attention at retrieval has not been evaluated.
The foregoing analyses suggest that some types of priming (e.g., conceptual or production) may rely on attention during retrieval. Other lines of research imply that implicit retrieval should be generally automatic (the automaticity hypothesis) (e.g., Gooding et al., 1999, Jacoby, 1991, Moscovitch, 1992). First, Logan (1990) argued that repetition priming and automaticity result from a common underlying mechanism—the storage and retrieval of representations of individual exposures to (or instances of) specific items. Logan argued that the re-exposure of an old item on an implicit test decreases the demands for processing that item. As a result, the improvement in performance (i.e., priming) is due to increased automaticity. Second, neuroimaging research supports the notion of increased automaticity in the processing of old items. Compared to new items, old items exhibit less neural activation, indicating that the priming component of implicit tests reflects a decrease in brain activity (a phenomenon referred to as repetition suppression; Grill-Spector, 2008, Horner and Henson, 2008). Finally, other analyses also imply that non-conscious retrieval is largely automatic (e.g., Jacoby, 1991, Voss and Paller, 2009, Yonelinas, 2002).
Prior research on attention and implicit retrieval is quite limited but has produced some support for the automaticity hypothesis (e.g., Clarke and Butler, 2008, Gooding et al., 1999, Helman and Berry, 2003, Jacoby et al., 1989; cf. Kinoshita, 1999). For example, Gooding et al. (1999) found no effect of DA at retrieval on associative priming. Participants in this study encoded pairs of words and were later tested with a word-stem completion task, using cue word—word stem pairs (e.g., dove—sti__). Distraction during the test did not diminish the amount of priming found. Similarly, Clarke and Butler (2008) found no reduction in word-stem completion priming under divided attention whereas word-stem cued recall was diminished. Kinoshita (1999) produced results partially consistent with these: in an associative priming task, divided attention did not affect priming for intact cue-target pairings (pairs that matched the study list), but reduced priming for re-arranged pairs in which an old target stem was paired with a cue word from a different study pair. Using related paradigms, Helman and Berry, 2003, Jacoby et al., 1989 found results consistent with the automaticity hypothesis. In an artificial-grammar learning task, Helman and Berry found that distraction during retrieval did not affect performance on the strings that were consistent with the learned grammar. Jacoby et al. found that the false fame effect (a misattribution in which the prior exposure of a non-famous name increases later fame ratings) was undiminished by divided attention during the fame ratings.
Given the paucity of prior research, it is not surprising that many of the critical issues regarding both implicit memory, and attention and memory retrieval, have not been addressed. Only one of the prior studies (Clarke & Butler, 2008) made use of standard repetition priming tasks. Furthermore, prior research did not explore different types of priming tasks (e.g., perceptual vs. conceptual, identification vs. production), known from the implicit memory literature to influence the nature of priming, nor different types of secondary tasks (e.g., material general vs. material specific), known from the attention and explicit memory literature to influence aspects of memory retrieval. Finally, the prior research generally did not provide information regarding secondary task costs (with the exception of Clarke and Butler, discussed later). This is critical because the automaticity hypothesis cannot be fully evaluated until costs (or lack thereof) are assessed in both memory and secondary task performance (e.g., Craik et al., 1996).
Section snippets
The present study
The present study has four goals: (1) broadly assess the effect of DA during implicit retrieval; (2) examine effects of DA across different types of implicit tests; (3) examine the effects of different types of secondary tasks in producing DA effects; and (4) assess the secondary task costs produced by implicit memory retrieval.
Some prior research stresses the automaticity of implicit retrieval (the automaticity hypothesis) which implies that implicit memory is generally unaffected by DA during
Experiment 1
The first experiment used a perceptual identification test. Attention was divided during the test phase only (all learning occurred under full attention). Because the test is perceptual and requires stimulus identification, both the perceptual-conceptual distinction and the identification-production distinction imply that it is the least likely type of test to be disrupted by DA. Despite this, prior research suggests that specific types of secondary tasks (e.g., those that have a material
Experiment 2
The first experiment found no negative effects of DA on priming in the perceptual identification task. However, this task is both perceptual and an identification task. Standard accounts of implicit memory imply that it is the form of implicit memory least likely to be affected by divided attention. To continue the investigation, the second experiment used the word-stem completion test. This test is classified as a perceptual task according to the TAP criteria and as a production task according
Experiment 3
The third experiment assessed a form of priming that should be most sensitive to attention according to standard theoretical analyses. This experiment used the category-exemplar production test, a test classified as conceptual by the TAP account and as a production task by the identification-production account. The former account predicts that conceptual tasks should be more sensitive to DA at retrieval than perceptual tasks, whereas the latter account predicts that production tasks should be
Experiment 4
The first three experiments indicate no effect of DA on priming in the implicit tests of perceptual identification, word-stem completion, and category-exemplar production. Prior research indicates that DA sometimes reduces explicit memory but does not always do so (e.g., Baddeley et al., 1984, Clarke and Butler, 2008, Craik et al., 1996, Hicks and Marsh, 2000, Lozito and Mulligan, 2006). Consequently, it is important to determine if the DA tasks used in the present experiments are capable of
Power analyses combining data across experiments
Our primary focus is on the effects of DA on priming. Experiment 4 provides reassuring evidence that the present DA manipulations are generally stringent enough to reduce explicit memory. However, because many of the conclusions about priming are based on null effects, power must be considered. To increase power, we conducted several further analyses, combining priming data across Experiments 1 through 3. The first analysis was a one-factor ANOVA, with 5 levels (FA and the four DA conditions).
General discussion
The priming results were quite consistent across Experiments 1–3. Despite substantial priming in the FA condition, dividing attention during retrieval produced no measurable effect on implicit memory. This equality of priming held across all four DA conditions, implying that the precise nature of the distracter task, material-specific vs. material-general tasks or high vs. low response frequency, was unimportant. Critically, explicit recall (in the word-stem cued recall test) was significantly
References (48)
- et al.
Brain regions associated with successful and unsuccessful retrieval of verbal episodic memory as revealed by divided attention
Neuropsychologia
(2005) Visual priming
- et al.
Priming, response learning and repetition suppression
Neuropsychologia
(2008) A process dissociation framework: Separating automatic from intentional uses of memory
Journal of Memory and Language
(1991)Repetition priming and automaticity: Common underlying mechanisms?
Cognitive Psychology
(1990)Attention and memory
- et al.
Category norms: An updated and expanded version of the battig and montague (1969) norms
Journal of Memory and Language
(2004) The nature of recollection and familiarity: 30 years of research
Journal of Memory and Language
(2002)- et al.
The attentional demands of encoding and retrieval in younger and older adults: I. Evidence from divided attention costs
Psychology and Aging
(1998) - et al.
Attention and retrieval from log-term memory
Journal of Experimental Psychology: General
(1984)
Category norms of verbal items in 56 categories: A replication and extension of the Connecticut category norms
Journal of Experimental Psychology Monographs
Attentional limits in memory retrieval
Journal of Experimental Psychology: Learning, Memory, and Cognition
Dissociating word stem completion and cued recall as a function of divided attention at retrieval
Memory
Statistical power analysis for the behavioral sciences
The effects of divided attention on encoding and retrieval processes in human memory
Journal of Experimental Psychology: General
Divided attention during encoding and retrieval: Differential control effects
Journal of Experimental Psychology: Learning, Memory, and Cognition
Divided attention and memory: Evidence of substantial interference effects at retrieval and encoding
Journal of Experimental: General
Interference effects from divided attention during retrieval in younger and older adults
Psychology and Aging
Impaired production priming and intact identification priming in Alzheimer’s disease
Journal of the International Neuropsychological Society
Convergent behavioral and neuropsychological evidence for a distinction between identification and production forms of priming
Journal of Experimental Psychology: General
Do novel associative word stem completion and cued recall share the same memory retrieval processes
Memory
Effects of divided attention and speeded responding on implicit and explicit retrieval of artificial grammar knowledge
Memory & Cognition
Towards specifying the attentional demands of recognition memory
Journal of Experimental Psychology: Learning, Memory, and Cognition
Category-item frequency and category name meaningfulness (m′): Taxonomic norms for 84 categories
Psychonomic Monograph Supplements
Cited by (29)
Implicit Versus Explicit Memory
2017, Learning and Memory: A Comprehensive ReferenceA Framework of Episodic Updating: An Account of Memory Updating After Retrieval
2017, Psychology of Learning and Motivation - Advances in Research and TheoryCitation Excerpt :Divided attention conditions involve performing a primary memory task and a secondary task simultaneously. In general, divided attention during retrieval has less pronounced effects on primary task performance than divided attention at encoding (Baddeley, Lewis, Eldridge, & Thomson, 1984; Craik et al., 1996; Lozito & Mulligan, 2010; Naveh-Benjamin, Craik, Guez, & Dori, 1998), though there are exceptions (e.g., Fernandes & Moscovitch, 2000; Hicks & Marsh, 2000; Lozito & Mulligan, 2006). One explanation of the distinct effects of divided attention on encoding and retrieval is that encoding allows more flexible control of attentional resources than does the retrieval process (Craik et al., 1996; Naveh-Benjamin, Craik, Perretta, & Tonev, 2000).
Attention and repetition priming in the verb generation task
2013, Acta PsychologicaCitation Excerpt :Third, the attentional demands of verb generation priming at retrieval are currently unknown. Recent studies have reported that priming in many tasks are unaffected by retrieval–phase DA, indicating that implicit memory retrieval processes may be automatic (Clark & Butler, 2008; Lozito & Mulligan, 2010). Whether that interpretation applies to the verb generation task is not yet known.
Conscious processing during retrieval can occur in early and late visual regions
2013, NeuropsychologiaCitation Excerpt :Explicit memory involves the conscious retrieval of previous experiences, whereas implicit memory involves nonconsious retrieval (e.g., Schacter, 1987; Lozito & Mulligan, 2010).
Long-term memory for distractors: Effects of involuntary attention from working memory
2024, Memory and Cognition