Developmental Psychology 2015, Vol. 51, No. 6, 731–743

© 2015 American Psychological Association 0012-1649/15/$12.00 http://dx.doi.org/10.1037/a0039057

The Interplay of Spatial Attentional Biases and Mental Codes in VSTM: Developmentally Informed Hypotheses Andria Shimi and Gaia Scerif

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

University of Oxford What cognitive processes influence how well we maintain information in visual short-term memory (VSTM)? We used a developmentally informed design to delve into the interplay of top-down spatial biases with the nature of the internal memory codes, motivated by documented changes for both factors over childhood. Seven-year-olds, 11-year-olds, and adults completed a VSTM task in which they decided whether a probe item had been present in a preceding memory array. Spatial cues guided participants’ attention to the likely location of the to-be-probed item during maintenance. We manipulated the memoranda to contain either highly familiar items or unfamiliar abstract shapes. All participants benefited from cues during maintenance, although benefits were smaller for 7-year-olds than for older participants. Critically, attentional benefits interacted with the nature of the memoranda: Better VSTM maintenance was obtained for cued familiar items. Furthermore, attentional benefits for familiar items correlated with validated measures of visual, but not verbal, short-term and working memory span. These data demonstrate that, in addition to the efficiency with which top-down biases operate during maintenance, the available mental codes for to-be-remembered items influence VSTM and differentially so over childhood. Attentional biases during maintenance seem to operate more efficiently on mental representations that are more robust and can be retrieved more easily. More important, this interaction follows a quantitative development. The findings elucidate further the dynamic interplay between attentional control and VSTM across development. Keywords: attentional orienting, visual STM, working memory, mental codes, development

Raye, Johnson, Mitchell, Reeder, & Greene, 2002). Finally, Awh and colleagues have suggested that attention-based spatial rehearsal mediates the maintenance of information in VSTM in a similar way as verbal rehearsal does for verbal information (Awh & Jonides, 2001; Awh, Jonides, & Reuter-Lorenz, 1998). VSTM capacity increases dramatically from childhood into adulthood (Cowan et al., 2005) and from early into later infancy (Ross-Sheehy, Oakes, & Luck, 2003). Driven by these advances in adult cognitive neuroscience, a wealth of recent studies have also examined the contribution of attentional biases and resources to changes in memory capacity (Burnett Heyes, Zokaei, van der Staaij, Bays, & Husain, 2012; Cowan, AuBuchon, Gilchrist, Ricker, & Saults, 2011; Markant & Amso, 2013; Ross-Sheehy, Oakes, & Luck, 2011). A recent line of developmental research has highlighted the distinction of multiple atttentional processes in service of VSTM: although adults’ attentional orienting in preparation for encoding is just as efficient as orienting during maintenance, research with children illustrated that these processes can fractionate, both in cognitive (Astle, Nobre, & Scerif, 2012; Shimi, Nobre, Astle, & Scerif, 2014) and neural terms (Shimi, Kuo, Astle, Nobre, & Scerif, 2014). For example, Shimi, Nobre et al. (2014) showed that while 7-year-old children benefited from cues presented before encoding to the same degree as 11-year-olds and young adults, their ability to use attention cues during maintenance was not as efficient as that of older individuals. Furthermore, children’s cueing benefits during maintenance predicted VSTM and visual working memory (VWM) span over and above differences in age and basic memory, suggesting that traditional span tasks and spatial attentional orienting during maintenance, but not before encoding, share com-

It is well established that visual short-term memory (VSTM) capacity is highly limited (Cowan, 2001; Luck & Vogel, 1997; Todd & Marois, 2004); therefore, it is essential that we select for encoding and maintenance only the information that is relevant to current task goals. Indeed, top-down attentional mechanisms have been proposed to facilitate both the encoding and maintenance of visuospatial material in adults. For example, Nobre and colleagues demonstrated that orienting attention via cues presented during the maintenance period optimizes VSTM (Griffin & Nobre, 2003; Kuo, Stokes, & Nobre, 2012; Nobre, Griffin, & Rao, 2007), by enhancing the active maintenance of relevant items and/or suppressing the activation of irrelevant items (Lepsien & Nobre, 2006). Similarly, Raye and colleagues have shown that attentionally refreshing a just-presented item by briefly thinking about it, resuscitates, strengthens, or prolongs its representation during maintenance (Raye, Johnson, Mitchell, Greene, & Johnson, 2007;

This article was published Online First April 6, 2015. Andria Shimi and Gaia Scerif, Department of Experimental Psychology, University of Oxford. Andria Shimi was supported by a Bodossaki Foundation scholarship, St. Peter’s College, University of Oxford, and by an A.G Leventis Foundation scholarship. Gaia Scerif was supported by a Scholar Award of the James S. McDonnell Foundation. We thank Elvira Masoura for providing us with the Greek versions of the two verbal subscales of AWMA. Correspondence concerning this article should be addressed to Andria Shimi, Department of Experimental Psychology, University of Oxford, South Parks Road, Oxford OX1 3UD, United Kingdom. E-mail: andria.shimi@ psy.ox.ac.uk 731

SHIMI AND SCERIF

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

732

mon processing demands. As a whole, this growing literature demonstrates that developing systems hold the promise of providing mechanistic insights into how spatial attentional biases facilitate memory performance, making them informative not only to researchers interested in the development of attention and VSTM (Cowan, Morey, AuBuchon, Zwilling, & Gilchrist, 2010; Kwon, Luck, & Oakes, 2014; Markant & Amso, 2014) but also to those studying these interactions in adults alone (Awh, Vogel, & Oh, 2006; Chun, Golomb, & Turk-Browne, 2011; Gazzaley & Nobre, 2012). While recent findings pointed to the developing ability to apply top-down spatial biases during maintenance as a critical factor for understanding the mechanisms underlying the development of VSTM, changes in attentional control alone cannot explain all developmental differences in VSTM capacity (Astle & Scerif, 2011). Indeed, a number of adult studies have shown that the content of VWM also drives attentional effects (Downing, 2000; Kumar, Soto, & Humphreys, 2009; Olivers, Meijer, & Theeuwes, 2006; Soto, Hodsoll, Rotshtein, & Humphreys, 2008; Woodman, Luck, & Schall, 2007). Johnson, Raye, Mitchell, Greene, and Anderson (2003), for example, showed that the distribution of hemodynamic brain activity in frontal areas of the brain during attentional refreshment of line drawings of familiar nameable objects, abstract patterns, and written words varied as a function of stimulus type, suggesting that the type of stimulus that is refreshed determines the recruitment of segregated neural circuits to facilitate ongoing cognition. Furthermore, Ricker and Cowan (2010) found differential effects on the maintenance of visual unconventional stimuli and familiar English letters and proposed that some information cannot go through attentional refreshment. In light of these findings, a critical question on what factors influence the efficiency of attentional orienting within VSTM over development pertains to the nature of the information to be maintained in memory itself. While the adult literature suggests a bidirectional relation between attention and WM content, this has been investigated through paradigms that involve searching displays for a specific memory representation and/or feature, examining the effects of memory representations upon visual search (Dombrowe, Olivers, & Mieke, 2010; Downing, 2000; Soto, Heinke, Humphreys, & Blanco, 2005). This has left, largely unknown, even in adults, the precise effects that VWM content has on top-down spatial biases during VWM maintenance. Could it be that the hitherto established top-down attentional biases of VSTM representations in adults (Stokes & Nobre, 2011) are further facilitated or constrained by the specific nature of the memoranda? To our knowledge, no study so far has examined the effects of differential memoranda on adults’ spatial attentional biases during maintenance. Furthermore, if they manifest in adulthood, are there age-related differences in the differential effects that the memoranda exert on top-down spatial modulations on VSTM representations? This is of theoretical value, as in a parallel vast literature, the nature of the internal memory codes afforded by different types of memoranda has been deemed critical for understanding developmental changes in VSTM (e.g., Cowan & Kail, 1996; a point to which we turn next).

Mental Codes and VSTM Maintenance Over Development Early influential cognitive theories incorporated the notion of multiple coding to explain differences in memory performance for

one set of material over the other. Age-related differences in VSTM were also explained within this theoretical framework. For example, the dual coding theory (Paivio, 1971) proposed two cognitive systems, one dedicated to processing and representation of nonverbal stimuli and one dedicated to verbal items. Dual coding would allow adults to create both visual and languagebased memory traces for nonverbal stimuli, leading to an additive benefit in recall; that is, representing an object in both systems (by generating multiple codes for it) is superior to representing the same object in only one system (Paivio & Csapo, 1973). Postle and colleagues documented the recruitment of multiple codes, both visual and verbal (with the latter possibly being semantic in nature), for the short-term retention of objects (Postle, D’Esposito, & Corkin, 2005). Similarly, according to the multicomponent model of WM, the episodic buffer provides visually presented material a gateway to semantic and LTM representations, signifying that a key function of WM is the feeding of information from and to LTM (e.g., Baddeley, 2000). Based on these accounts, memory codes may be boosted for one set of material over the other by knowledge of the items to be remembered. This is a factor that has already been suggested to contribute to memory performance (Cowan, 1997; Cowan & Alloway, 2009) presumably because it grants faster access to semantic codes (Potter & Faulconer, 1975). However, it has been studied substantially less in the visual compared with the verbal domain (Pickering, 2001). Understanding the effects of multiple coding for the visual domain is important because visuospatial stimuli differ in the extent to which they afford multiple mental codes during maintenance (Farmer, Berman, & Fletcher, 1986; Kemps, De Rammelaere, & Desmet, 2000; Pickering, Gathercole, Hall, & Lloyd, 2001; Ricker, Cowan, & Morey, 2010; Vandierendonck, Kemps, Fastame, & Szmalec, 2004). Qualitative changes in the ability to use multiple mental codes were initially suggested to account for the fact that visuospatial memory improves significantly over childhood. For example, studies that examined immediate serial recall of familiar, nameable pictorial items demonstrated that children in late childhood and adults have a pervasive tendency to recode such items into verbal form, unless they are prevented from doing so (e.g., Brandimonte, Hitch, & Bishop, 1992; Hitch, Halliday, Schaafstal, & Heffernan, 1991; Hitch, Halliday, Schaafstal, & Schraagen, 1988; Hitch, Woodin, & Baker, 1989; Longoni & Scalisi, 1994; Palmer, 2000; Schiano & Watkins, 1981), whereas young children spontaneously rely on visual codes to maintain information until the age of 7– 8 (Gathercole, 1998; Gathercole & Hitch, 1993; Hitch, Halliday, Hulme, Voi, Routh, & Conway, 1983). However, young children have been found to rely on verbal codes when required to retain auditory and spoken items (e.g., Hitch, Halliday, Dodd, & Littler, 1989; Hulme, Thomson, Muir, & Lawrence, 1984) and other adult findings showed that visual codes are not entirely lost over development (Della Sala, Logie, Marchetti, & Wynn, 1991; Logie, Del Sala, Wynn, & Baddeley, 2000; Saito, Logie, Morita, & Law, 2008). These influential developmental studies examined primarily the role of phonological recoding and/or rehearsal in VSTM, whereas more recent adult studies have examined the multiplicity of mental codes entailed in VWM (e.g., Postle et al., 2005). However, taken together, these findings indicate that differences in VSTM across age groups reflect at least in part changes in the ability to flexibly access and maintain multiple appropriate mental

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

ATTENTIONAL BIASES AND THE NATURE OF MEMORANDA

codes if the visuospatial task at hand allows it (Hitch, 2006; Kwon, Reiss, & Menon, 2002; Palmer, 2000). Nevertheless, changes in the ability to use multiple codes alone cannot account for all developmental differences in VSTM (Pickering, 2001). As discussed earlier, internal attentional orienting (Griffin & Nobre, 2003; Kuo et al., 2012) and attentional refreshment (Raye et al., 2007) have been proposed to contribute to the maintenance of visuospatial material in adults. It is plausible then that developmental differences in VSTM performance capture the interaction of the two processes. That is, they may represent older individuals’ developed ability to strategically use the most appropriate mental codes on the one hand, which has a benefit on retaining enhanced visual representations internally (Paivio, 1971), and their refined ability to apply spatial biases to maintain a selected internal representation on the other hand (Shimi, Nobre, et al., 2014). Therefore, studying attentional benefits on maintenance and the benefits for memoranda that afford multiple mental codes within a developmentally informed design provide the unique opportunity to test the interplay between topdown spatial attentional biases and internal mnemonic information. To our knowledge, no study so far has examined whether the pattern of attentional biases during VSTM maintenance differs across development as a function of the type of items that it needs to operate on.

Hypotheses This study examined whether participants’ proficiency in maintaining information in VSTM depends on the interaction between the abilities to deploy spatial attentional biases and to represent objects by pooling multiple mental codes. Multiple codes could facilitate VSTM because when an attention cue appears during the maintenance period participants can access and reactivate the representation of the selected item more reliably. Young children may be less able to access and selectively refresh a representation held in VSTM because their ability to use such coding is less developed. To investigate this hypothesis, we contrasted memory for highly familiar and easy to name items that are traditionally learned very early in life (i.e., animals) with that for novel meaningless shapes that are difficult to label. We hypothesized that if adults and older children pervasively generate multiple codes to represent visual pictorial material to facilitate maintenance of information until response, in addition to the greater ability to use cues during maintenance, they should perform significantly better than younger children in a cued condition with highly familiar items than in a cued condition with unfamiliar items. If young children are perceptually bound and spontaneously rely only on visual codes for visually presented material, despite the availability of additional nonvisual codes,1 then they should perform similarly across item types. Conversely, if participants across age groups rely on visual features and the only factor that influences performance is the ability to recruit attentional control during maintenance, then all participants should perform equally across familiar and unfamiliar stimuli, although older participants should perform better in cued trials than young children because of better attentional control (Astle et al., 2012; Shimi, Nobre, et al., 2014) and in neutral trials because of larger storage capacity (cf. Cowan, AuBuchon, Gilchrist, Ricker, &

733

Saults, 2011; Cowan, Morey, AuBuchon, Zwilling, & Gilchrist, 2010).

Method Participants Twenty children (10 boys and 10 girls) aged 6 and 7 years old (M ⫽ 6.50 years old, SD ⫽ .51; “7-year-olds” henceforth), 20 children (9 boys and 11 girls) aged 10 and 11 years old (M ⫽ 10.5 years old, SD ⫽ .51; “11-year-olds” henceforth), and 20 adults (8 males and 12 females) between 21 and 27 years old (M ⫽ 23.90 years old, SD ⫽ 1.86) participated in the study. Children were recruited from local primary schools via an opt-in procedure. Adults were recruited among university undergraduates and postgraduates. Children with neurological/psychiatric disorders were excluded from the study and no hearing problems were reported by the parents or teachers of the participants. All participants were healthy and had normal or corrected-to-normal vision. Before testing, ethical approval from the appropriate Research Ethics Board was secured and permission for entry in schools was given by relevant local Educational Boards. Parents/guardians of children participants and adult participants signed informed-consent forms and children also verbally assented to participate in the study.

Apparatus Attentional orienting task. We used an attentional orienting task in which we manipulated the items in the memory array to be either familiar nameable items or meaningless difficult-to-label shapes (see detailed description of items below). The task is presented schematically in Figure 1A. In each trial participants viewed briefly a memory array of four peripheral items, followed by a single central probe item after a variable delay (see Figure 1A) and were instructed to indicate whether the probe was present in the preceding memory array by pressing the mouse buttons. Memory arrays were always followed by attention cues that were spatially informative or uninformative to the location of the item to be probed, depending on the trial type: retro-cue or neutral (see Figure 1B). In retro-cue trials, a black arrow (spatially informative cue) was presented after the memory array and guided participants’ attention internally to the mental representation of one of the already encoded items of the array. Retro-cues were always valid (100% validity) in directing attention to the location of the item to be probed. In neutral trials, a black filled square (spatially uninformative cue) was presented after the memory array. The black filled square was presented to control for generic alerting effects. Memory array items were black and white line drawings of animals (“familiar” items henceforth) or abstract shapes (“unfamiliar” items henceforth; see Figure 1B) and were presented on a white background with E-Prime 1.2 (Psychological Software Tools, Inc., Sharpsburg, PA) on a PC screen. They were symmetrically arranged, each cen1 It should be noted that although previous studies typically assume that nonvisual codes are phonological (i.e., auditory or speech-based), it is possible that they are semantic in nature. Therefore, when using the term “language-based” we refer to nonvisual codes as these may or may not denote codes that are phonological in nature.

SHIMI AND SCERIF

734 A

ISI between array and cue 800-1200 ms

+

500 ms

+

500 ms 350 ms

ISI between cue and probe 800-1200 ms

+

300 ms

+

+

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

350 ms

The interplay of spatial attentional biases and mental codes in VSTM: Developmentally informed hypotheses.

What cognitive processes influence how well we maintain information in visual short-term memory (VSTM)? We used a developmentally informed design to d...
323KB Sizes 0 Downloads 7 Views