حواس پرتی چند وجهی: بینش توجه محدود کودکان
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|38769||2015||10 صفحه PDF||سفارش دهید||7348 کلمه|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Cognition, Volume 136, March 2015, Pages 156–165
Abstract How does the multi-sensory nature of stimuli influence information processing? Cognitive systems with limited selective attention can elucidate these processes. Six-year-olds, 11-year-olds and 20-year-olds engaged in a visual search task that required them to detect a pre-defined coloured shape under conditions of low or high visual perceptual load. On each trial, a peripheral distractor that could be either compatible or incompatible with the current target colour was presented either visually, auditorily or audiovisually. Unlike unimodal distractors, audiovisual distractors elicited reliable compatibility effects across the two levels of load in adults and in the older children, but high visual load significantly reduced distraction for all children, especially the youngest participants. This study provides the first demonstration that multi-sensory distraction has powerful effects on selective attention: Adults and older children alike allocate attention to potentially relevant information across multiple senses. However, poorer attentional resources can, paradoxically, shield the youngest children from the deleterious effects of multi-sensory distraction. Furthermore, we highlight how developmental research can enrich the understanding of distinct mechanisms controlling adult selective attention in multi-sensory environments.
1. Introduction The effectiveness of cognitive functioning in everyday life is determined by the ability to focus on a task while ignoring concurrent distracting stimuli (i.e., selective attention). Models of attentional selection were greatly advanced by “perceptual load theory” (e.g., Lavie, 1995, Lavie, 2005, Lavie, 2010 and Lavie and Tsal, 1994), proposing that the extent to which such irrelevant stimuli are distracting is determined by the degree to which the currently performed task exhausts one’s available attentional resources. This influential proposal operationalized “distraction” as interference on one’s primary task by task-irrelevant stimuli and we shall here follow this convention. The current study demonstrates that studying distraction in real-life environments, multi-sensory by nature, can reveal other mechanisms important for controlling attention, and that their importance might be more readily witnessed by studying cognitive systems whose attentional control is developing (e.g., children). 1.1. Attentional allocation in unimodal environments Lavie and Tsal (1994) argued that attentional resources, in particular their limited nature, are what determines whether stimuli irrelevant to the current task will be processed. Lavie and colleagues provided evidence for this claim in a series of now classical studies that employed the response-competition task (Lavie, 1995, Lavie and Cox, 1997 and Lavie and Tsal, 1994): Typically, when one is searching for one of two target letters (X or N) amongst a small number of letters (a task posing low perceptual load demands), concurrently presented peripheral distractors trigger reliable stimulus–response compatibility effects, i.e., slower search times on trials in which these peripheral stimuli prime a response opposite to the target response (e.g., an X when the target was an N). However, during search amongst a larger number of similar letters (a task posing higher perceptual load demands) compatibility effects are strongly reduced. In line with perceptual load theory (Lavie, 1995), in a task posing low perceptual demands, remaining attentional resources are automatically allocated to task-irrelevant stimuli in the environment. This results in distraction, as both target and distractors are processed up to the stage of their semantic representation and associated motor response. Such a situation contrasts with processing of distractors in a task that is perceptually demanding: Their processing is reduced or even eliminated, because the task is thought to be exhausting the available attentional resources. 1.2. Attentional allocation when faced with cross-modal distraction While the importance of the nature of one’s primary task in constraining distraction has since been replicated with various methods, measures and populations (see Lavie, 2010, for a review), of particular value is testing whether predictions of perceptual load theory hold against everyday situations, such as in the context of cross-modal distraction. Early seminal work by Allport and colleagues (e.g., Allport, Antonis, & Reynolds, 1972) had demonstrated that a fairly complex auditory task (i.e., auditory shadowing) can be performed alongside a demanding visual task (i.e., sight-reading music), which suggests a limited effect of processing load across senses. Further contrasting evidence was provided by Tellinghuisen and Nowak (2003), who used a version of the response-competition task adapted to a cross-modal context: When peripheral letter distractors are presented auditorily during search for visual letter targets, they, unlike visual distractors, filter into further processing stages, causing interference under conditions of high visual perceptual load. The residual interference effects from auditory distractors on visual tasks have been presented as evidence for separate attentional resources in vision and audition. Visual distractors do not impact attention on the primary task, presumably because attentional resources in the primary modality have been depleted, whereas separate resources are devoted to auditory distractors (Duncan et al., 1997 and Welch and Warren, 1980). However, recent studies have provided mixed evidence for this account (Jacoby et al., 2012 and Klemen et al., 2009; Parks, Hilimire, & Corballis, 2011). For example, high visual perceptual load was recently shown to induce inattentional deafness: Macdonald and Lavie (2011, Experiment 3) instructed participants to judge which of two coloured arms of a centrally presented cross was longer, while on some trials a task-irrelevant pure tone was presented. On trials where the two arms differed in length only slightly (a perceptually demanding task), conscious awareness of the tone was reduced compared to trials in which the difference in arm length was larger (a task with lower perceptual demands). In contrast to separate-resources models, these results indicate that in adults, even in cross-modal contexts, at least under some conditions (e.g., very high visual load and/or complete task irrelevance of the auditory distractor) attentional resources are shared across modalities. 1.3. Attentional allocation in multi-sensory environments The jury is therefore still out on whether cross-modal distraction can be entirely removed by increases in visual attentional load and on what drives cross-modal distraction, i.e., interference, on a visual task. Particularly informative to this debate are studies employing stimuli that present redundant information to more than one modality at once (e.g., Matusz and Eimer, 2011, Matusz and Eimer, 2013 and Van der Burg et al., 2011). Multiple sources of congruent information are integrated into a unified multi-sensory percept that triggers enhanced behavioural and/or neural responses, both when the information is redundant at a low perceptual (e.g., temporal and/or spatial alignment; e.g., Santangelo & Spence, 2007, but see Spence, 2010) or high semantic level (e.g., Laurienti, Kraft, Maldjian, Burdette, & Wallace, 2004; for a review, see Alais, Newell, & Mamassian, 2010). However, this body of research has tended not to use the classical visual perceptual load paradigms. Yet, this novel extension is much needed, as it would bridge the perceptual load theory of selective attention and theories of multi-sensory processing, which traditionally have been developed separately. Do increased perceptual demands of the primary task reduce distraction elicited by multi-sensory events? If audiovisual distractors were processed under both lower and higher visual load, this would provide further support for the idea that, at least under some conditions, separate attentional resources are deployed (Tellinghuisen & Nowak, 2003). Interestingly, multi-sensory distractors should generally result in more robust distraction (i.e., interference on the primary task) than unimodal distractors because at each level of visual perceptual load they would engage attentional resources in two modalities. If such effects were indeed observed, this would call for a revision of the perceptual load theory to accommodate multi-sensory distraction. 1.4. Insights from developmentally-informed research Some of the strongest evidence for the critical role of attentional resources in reducing distractor processing has been provided by research involving young children, whose attention is known to be less efficient than that of adults (e.g., Plude et al., 1994 and Trick and Enns, 1998). In a version of the response-competition task, Huang-Pollock, Carr, and Nigg (2002) found that children as young as seven years of age were more distracted by peripherally-presented letters than young adults when the search task was easy, consistent with poorer mechanisms of distractor interference control in children (Posner, Rothbart, & Thomas-Thrapp, 1997). Under conditions of high perceptual load, children were not differentially influenced by distractors, like adults, as indexed by a lack of distractor compatibility effects on RTs (see also Maylor & Lavie, 1998, for similar implications from data from elderly participants). This pattern of results suggested that children are less able than adults to control stimulus–response conflict (a marker of poor attentional control), but only until attentional-capacity limits are reached. A developmental approach might be beneficial also in assessing whether the predictions of perceptual load theory extend to how attentional resources are allocated to distractors presented in multiple sensory modalities. Do increased perceptual demands of the primary task reduce distraction elicited by multi-sensory distractors in children, who have fewer attentional resources, as well as in adults? A developmentally-informed design therefore has the potential to provide insights into distinctive mechanisms controlling attention in multi-sensory contexts. Critically, while children and adults seem to be similarly ‘shielded’ from visual distraction at higher levels of visual load, this might not hold true for distractors in other modalities, and especially ones presenting information to multiple sensory modalities at once. In systems possessing weak attentional resources, the perceptual load theory in its current form would expect higher levels of visual load to exhaust these resources earlier, thus decreasing the processing of multi-sensory distractors compared to the fully developed system. However, if interference from multi-sensory distractors was found for children, this would call for a revision of perceptual load theory to accommodate the role of multi-sensory distraction. One would need to account for how increases in the perceptual load of the primary task may shield from distraction under some conditions but not others. These are as yet untested hypotheses, despite their clear importance for selective attention models and for the understanding of attentional control development. 1.5. The current study: approach and predictions The aim of this study was to investigate developmental differences in how the perceptual load of a primary visual selective attention task constrains the processing of multi-sensory, i.e., audiovisual, distractors. For this purpose, we employed a traditional perceptual load paradigm, with a novel modification: For the first time, peripheral distractor stimuli were not only presented visually or auditorily, but also audiovisually. Secondly, and again for the first time, age-related differences in attentional abilities were used to probe the limits of multi-sensory distraction. Six-, 11- and 20-year-olds searched for a visual coloured shape (red square or green circle) in arrays consisting of 1 (a lower visual load condition) or 4 (a higher visual load condition) coloured shapes. For visual distractors, at the lower level of visual load we predicted larger compatibility effects for children than adults, because of children’s poorer control of stimulus–response conflict, an attentional control mechanism (replicating Huang-Pollock et al., 2002). At the higher level of load, the exhaustion of visual attentional resources should eliminate visually-induced compatibility effects across all age groups. For auditory distractors, compatibility effects in adults were expected not to be modulated by visual load (Tellinghuisen & Nowak, 2003), although these findings remain controversial (Macdonald & Lavie, 2011). Critically, in adults we expected robust distraction in response to audiovisual distractors, i.e., compatibility effects at levels of lower and higher load that, at a minimum, resemble the processing of the most effective distractor at each level. A developmentally-inspired design provided informative differential predictions with respect to distinct mechanisms controlling distraction in multi-sensory contexts: if attentional resources are joint across modalities (Macdonald & Lavie, 2011), auditorily – and audiovisually-induced compatibility effects should be eliminated or at least strongly reduced across all age-groups at the higher level of visual load. If instead separate attentional resources exist for the visual and auditory modality, audiovisually-induced compatibility effects should not be reduced by visual load, even in the youngest children.
نتیجه گیری انگلیسی
3. Results Means of median correct reaction time (RT) and error rates are reported in Fig. 2. A four-way mixed analysis of variance (ANOVA) was conducted on the RTs data, with compatibility (distractor compatible versus incompatible with the target identity), set-size (1 versus 4 coloured shapes in the search array), and distractor type (visual versus auditory versus audiovisual) as within-subjects factors, and age (adults versus 11-year-olds versus 6-year-olds) as a between-subjects factor. RTs were modulated by age, F(2,90) = 59.51, p < .001, ηp2 = .57, with gradually faster responses across age-groups (1265 ms versus 1070 ms versus 699 ms), all p’s < .001. There were main effects of compatibility, F(1,90) = 120.47, p < .001, ηp2 = .57; distractor type, F(1.83,164.92) = 16.1, p < .001, ηp2 = .15; and set-size, F(1,90) = 111.3, p < .001, ηp2 = .55. Compatibility and set-size interacted, F(1,90) = 24.41, p < .001, ηp2 = .21, and age modulated this interaction, F(2,90) = 7.48, p < .001, ηp2 = .14, suggesting that increasing set-size reduced distraction, but also that this effect differed across age groups. A three-way interaction between compatibility, set-size and distractor type, F(2,180) = 21.8, p < .001, ηp2 = .2, indicated that the effect of increased set-size on distraction also differed depending on the type of distractor. Median correct RTs and mean error rates observed for young adults (left panels), ... Fig. 2. Median correct RTs and mean error rates observed for young adults (left panels), 11-year olds (middle panels) and 6-year olds (right panels) on compatible and incompatible trials at two levels of set size, shown separately for visual, cross-modal and audiovisual distractors. The error bars represent standard error of the mean. Figure options Critically, a four-way interaction between compatibility, set-size, distractor type and age was observed, F(4,180) = 3.89, p < .01, ηp2 = .01. To investigate the sources of this interaction, compatibility effects were analysed separately for each distractor type. To summarize these results, while increased visual set-size removed interference effects of visual-only distractors in all age-groups, audio-visual distractors affected adults at both levels of set-size. For all children, increased visual set-size reduced audiovisual distraction significantly, and to the level of only a trend for the youngest children. In addition, we investigated the 4-way interaction effect by comparing the effects of distractor types directly. At set-size 1, for all groups, audio-visual and visual distractors resulted in larger interference effects compared to auditory distractors. At set-size 4, adults’ responses only were more affected by audio-visual than by auditory distractors. 3.1. Visual distractors Overall faster responses on compatible relative to incompatible trials (965 ms versus 1084 ms), F(1,90) = 53.44, p < .001, ηp2 = .37, were modulated by set-size, F(1,90) = 35.45, p < .001, ηp2 = .28. Pair-wise comparisons revealed reliable compatibility effects elicited at set-size 1 (228 ms, p < .001), but not set-size 4 (14 ms, p = .56). Importantly, compatibility effects were modulated by age, F(2,90) = 6.62, p < .001, ηp2 = .13. Here and henceforth, significant interactions were investigated with analyses of simple main effects. Pair-wise comparisons revealed that overall compatibility effects triggered by visual distractors in both younger (149 ms) and older (173 ms) children were reliably larger when compared to adults (36 ms), smaller p < .05, but no difference was found between the two groups of children, p = 1. As predicted, there was a three-way compatibility × set-size × age interaction, F(2,90) = 8.07, p < .001, ηp2 = .15, indicating that reductions of compatibility effects across set-sizes differed in adults and children (see Fig. 2, top panel). Pair-wise comparisons demonstrated that in all age groups reliable compatibility effects were elicited at set-size 1 (345 ms in 6-year-olds; 281 ms in 11-year-olds, and 53 ms in adults, all p’s < .001). Separate pair-wise comparisons demonstrated these compatibility effects elicited at set-size 1 were reliably larger in both younger and older children compared to adults, p’s < .001, but not different across the two groups of children, p = .83. Critically, at set-size 4, visually-induced compatibility effects were completely eliminated in all age groups (smallest p = .076). To investigate whether the observed pattern was due to generally slower responses in children, compatibility effects were scaled by average RTs across conditions (see Huang-Pollock et al., 2002 and Maylor and Lavie, 1998). An ANOVA on these proportional scores retained a set-size x age group interaction, F(2,90) = 3.57, p < .05, ηp2 = .07. As seen previously, for all groups significant compatibility effects (p’s < .001) at set-size 1 were eliminated at set-size 4 (p’s > .11), p’s < .05. 3.2. Auditory distractors RTs were overall reliably faster on compatible versus incompatible trials (946 ms versus 1005 ms), as indexed by a main effect of compatibility, F(1,90) = 11.42, p < .001, ηp2 = .12. As shown by Fig. 2 (middle panel), in contrast with the effects found for visual distractors, these compatibility effects were not modulated by age, F(2,90) = 1.45, p = .24, set-size, F(1,90) = 1.78, p = .19, or an interaction between set-size and age, F < 1. An anonymous reviewer helpfully pointed out that, although interaction effects did not reach statistical significance, visual inspection of Fig. 2 suggested that compatibility effects might not have been reliably triggered in adults by auditory distractors (see left column in the middle panel). Separate pair-wise comparisons confirmed this for set-size 1 (7 ms, p = .15), with distraction effects observed at set-size 4 (30 ms, p < .05). In 11-year-olds compatibility effects were reliably present both at set-size 1 (42 ms, p < .05) and set-size 4 (94 ms, p < .05), while in 6-year-olds they were significant at set-size 1 (79 ms, p < .05), but at a level of a non-significant trend at set-size 4 (103 ms, p = .06). To reiterate, the most conservative statistical analyses revealed no differences across these compatibility effects. Both this and the lack of reliable cross-modal distraction effects at lower levels of load in adults are consistent with previous work ( Tellinghuisen & Nowak, 2003). 3.3. Audiovisual distractors There were overall faster responses on compatible relative to incompatible trials (948 ms versus 1124 ms), F(1,90) = 98.69, p < .001, ηp2 = .52. This effect was modulated by set-size, F(1,90) = 19.95, p < .001, ηp2 = .18, with compatibility effects larger at set-size 1 than set-size 4 (243 ms versus 107 ms). Importantly, compatibility interacted with age, F(2,90) = 9.22, p < .001, ηp2 = .17, with larger compatibility effects when both younger (242 ms) and older (211 ms) children were compared to adults (68 ms), smaller p < .01, but with no difference between two groups of children, p = 1. Similarly to visual distractors, a three-way compatibility × set-size × age interaction was found, F(2,90) = 5.87, p < .01, ηp2 = .12, suggesting that the reduction of compatibility effects as a function of set-size differed again across ages (see Fig. 2, bottom panel). In an analysis of simple main effects, a first series of separate pair-wise comparisons revealed that the three-way interaction effect was driven by the fact that compatibility effects emerged at both levels of set-size for adults (68 ms and 68 ms, both p < .001), and for 11-year-olds (285 ms at set-size 1, while they were reduced at set-size 4, 136 ms, smaller p < .01), whereas for 6-year-olds they were reliable at set-size 1 (371 ms, p < .001), but not at set-size 4 (113 ms, p = .054). A further series of pair-wise comparisons carried out on these compatibility effects revealed that compatibility effects were reliably reduced between set-size 1 and 4 for both younger (t(29) = 3.41, p < .01) and older children (t(32) = 3.04, p < .01) , but not adults (t(29) = .01, p = .99). An ANOVA run on proportional compatibility effects (i.e., compatibility effects scaled by average RTs across conditions as above) for audiovisual distractors was carried out to compare them more fairly across age-groups, as these differed widely in average RT. The ANOVA retained a set-size × age group interaction, F(2,90) = 4, p < .05, ηp2 = .09. For all age groups audiovisual distraction effects were reliably present at set-size 1 (scaled compatibility effects were .271, .274, .109 for younger, older children and adults respectively, p’s < .001), while at set-size 4 they were eliminated for 6-year-olds (.070, p = .078), p < .01 for the decrease from set-size 1 to set-size 4, and attenuated for 11-year-olds (.109, p < .01), p < .001 for the decrease from set-size 1, but remained robust across this set-size for adults (.080, p < .001), p = .15 for the null decrease from set-size 1. 3.4. Differences across distractor types The interaction effects between distractor type and the other factors, reported above, called for an additional explicit comparison of distractor effects. We ran three separate repeated measures three-way ANOVAs, one for each age group of interest, with compatibility, set-size, and distractor type as within-subjects factors. For younger children, the interaction between distractor type, set-size and compatibility was significant, F(2,58) = 11.425, p < .001, ηp2 = .283, driven by simple main effects of distractor type for set-size 1, for both compatible, F(2,28) = 5.228, p = .011, ηp2 = .274, and incompatible trials, F(2,28) = 19.199, p < .001, ηp2 = .578, but not for set-size 4 (p > .305). At set-size 1, younger children responded significantly faster with auditory distractors than with visual and audiovisual distractors, for compatible and incompatible trials (p < .013 and p < .001, respectively), whereas visual and audiovisual distractors did not differ (p = .159). For older children, there was a reliable interaction between distractor type, set-size and compatibility, F(2,58) = 8.959, p < .001, ηp2 = .219, driven by a simple main effect of distractor type for incompatible trials at set-size 1, F(2,28) = 31.015, p < .001, ηp2 = .667. Older children were faster at responding with incompatible auditory than incompatible visual and audiovisual distractors (p < .001), whereas visual and audiovisual distractors did not differ (p > .922), and there were no distractor effects on compatible trials (p = .880). For adults, there was only a two-way interaction between distractor type and compatibility, F(2,58) = 8.177, p = .001, ηp2 = .220, driven by a simple main effect of distractor type for incompatible trials, F(2,28) = 14.865, p < .001, ηp2 = .515, but not compatible trials (p = .920). Adults were faster with incompatible auditory distractors (M = 695 ms) than with visual distractors (713 ms, p = .035) and were slowest with audio-visual distractors (747 ms, p = .001 compared to both auditory and audiovisual distractors). As this pattern may suggest evidence of enhanced multi-sensory compared to unimodal effects in adults, targeted pairwise comparisons on compatibility effects at each level of set-size were conducted. At set-size 1, for adults compatibility effects for audiovisual (68 ms) and visual distractors (i.e., the most effective unimodal distractor at that set-size, 53 ms) did not differ significantly, p = .310. At set-size 4, for adults the compatibility effect for audiovisual distractors (68 ms) was significantly larger than that for auditory distractors (i.e., the most effective unimodal distractor at that set-size, 30 ms, p = .024). 3.5. Accuracy data Error rates data failed to fulfil the parametric test assumptions. Wilcoxon’s signed rank tests showed visually-induced compatibility effects at set-size 1 and 4 for both 6-year-olds (20% and 10.3%, p’s < .001) and 11-year-olds (11.4% and 9.1%, p’s < .05), but not in 20-year-olds (1.8%, p = .13, and 3.1%, p = .053). Auditorily-induced effects were reliable only in 6-year-olds at set-size 1 (5.1%, p < .05), p’s > .27 for other groups and conditions. Audiovisually-induced effects were significant for 11-year-olds at set-size 1 and 4 (11.2% and 10.6%, p’s < .001), and significant at set-size 1 but not 4 for 6-year-olds (21.8%, p < .001, and 3.8%, p = .5) and 20-year-olds (2.7%, p < .03 and −.09%, p = .68), respectively. Despite the fact that the error rates were particularly high for some of the youngest children (see Fig. 2), the reduction in compatibility effects as measured by RTs with increased set-size was not due to a speed accuracy trade-off. The same pattern of results was revealed by an identical four-way ANOVA when participants with error rates above 33% (twelve 6-year-olds and two 11-year-olds) were excluded from the analyses, and thus these analyses are not reported here.