Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2021 Mar 1.
Published in final edited form as: J Exp Child Psychol. 2019 Nov 23;191:104732. doi: 10.1016/j.jecp.2019.104732

What Underlies Visual Selective Attention Development? Evidence that Age-related Improvements in Visual Feature Integration Influence Visual Selective Attention Performance

Andrew Lynn 1, Elena K Festa 1, William C Heindel 1, Dima Amso 1
PMCID: PMC6983333  NIHMSID: NIHMS1542609  PMID: 31770683

Abstract

Visual selective attention (VSA) improves across childhood. Conjunction search tasks require integrating multiple visual features in order to find a target among distractors and are often used to measure VSA. Motivated by the visual system’s architecture and developmental changes in neural connectivity, we predicted that feature integration across separate visual pathways (e.g., color and motion) should develop later than feature integration within the same visual pathways (e.g., luminance and motion). Eighty-nine 4- to 10-year-old children completed a visual search task that manipulated whether feature integration was between separate, parallel visual pathways or within the same visual pathway. We first examined whether color-motion integration was associated with a performance cost relative to luminance-motion integration across childhood. We found that color-motion integration was worse than luminance-motion integration in early childhood, but that this difference decreased with age. We also examined whether luminance-motion and color-motion and visual search performance developed differently across childhood. Reaction time (RT) visual search slopes for the luminance-motion condition were both stable across childhood and overall steeper than the color-motion condition. In contrast, RT search slopes for the color-motion condition became steeper across childhood. Finally, we found that age-related improvements in color-motion integration, relative to luminance-motion integration, was associated with longer color-motion search rates across childhood. These data suggest that age-related improvements in color-motion feature integration may increase competition between color-motion targets and distractors, thereby increasing the amount of time needed to process distractors as non-targets during the selection process.

Keywords: visual attention, selective attention, visual search, feature integration, development

Introduction

Visual selective attention (VSA), in which certain visual objects or locations are selected in the presence of competing others (Desimone & Duncan, 1995; Treisman & Gelade, 1980), typically improves during childhood, through adolescence, and peaks in early adulthood (e.g., Hommel, Li, & Li, 2004; Trick & Enns, 1998). VSA has been found to be a critical component of effective learning and memory in both infants (Markant, Ackerman, Nussenbaum, & Amso, 2016; Markant & Amso, 2013) and children (Markant & Amso, 2014). Yet, the mechanisms underlying the development of this key process are not well understood. Here, we ask whether age-related changes in visual feature integration shape VSA.

Visual search tasks, often used to study VSA, require participants to search for a target among competing distractors (e.g., Treisman & Gelade, 1980). Targets and distractors vary along one or more visual feature dimensions (e.g., color, orientation). During a “conjunction search,” a target defined by two or more visual features (e.g., a red bar oriented at 60°) is presented among distractors that share one value along one feature dimension, but differ in value along a second feature dimension (e.g., red bars oriented at 90° and green bars oriented at 60°). Thus, participants must integrate multiple visual features as they search amongst targets and distractors. Typically, the response time to find conjunction targets increases linearly as distractor number increases (RT slope), reflecting attentional engagement and visual search rate (e.g., Treisman & Gelade, 1980; Wolfe, 1994).

Developmental studies of VSA that employ visual search tasks reveal general improvements in processing speed, but also nuances in VSA as a function of task demands (Lobaugh, Cole, & Rovet, 1998; Trick & Enns, 1998). Beginning in infancy and toddlerhood, conjunction visual search performance shows patterns consistent with adult patterns in corresponding visual search tasks, but children’s search rates (RT slope) become faster across toddlerhood (Gerhardstein & Rovee-Collier, 2002). Similarly, studies have found that, while conjunction search rate for color-defined oriented bars was slower in children relative to adults, search rates became faster from middle (7 years) to late childhood (10 years) (Donnelly et al., 2007). Similarly, conjunction search rate for a luminance-defined shape (e.g., black circle) was slower in middle childhood relative to late childhood which was slower than in adulthood (Merrill & Lookadoo, 2004). However, search rates became adult-like by late childhood when researchers varied the amount of distractor competition by holding one distractor type constant (e.g., black square) while increasing only the second distractor type (e.g., grey circle). Here we asked whether developmental improvements in feature integration are an agent of change in conjunction visual search performance from early, across middle, and into late childhood (4–10 years).

Given that conjunction visual search requires integrating multiple visual features, it is important to consider that visual features are processed in a distributed set of hierarchically organized, parallel neural pathways (e.g., Felleman & Van Essen, 1991; Ungerleider & Haxby, 1994; Zeki, 1978). While some visual features are processed in relatively distinct pathways, others are processes within the same pathway. For example, color and motion information are processed in relatively distinct, but overlapping layers in cortical areas V1 and V2 and then routed to separate higher-level extrastriate cortical areas V4 and MT, respectively (Gegenfurtner, 2003; Seymour, Clifford, Logothetis, & Bartels, 2009; Shipp & Zeki, 1995; Sincich & Horton, 2005). However, luminance information proceeds with motion information along the visual hierarchy from V1, through V2, to MT. Thus, feature integration may occur across separate visual pathways (e.g., color and motion) or within the same visual pathway (e.g., luminance and motion). In this example, both across and within pathway feature integration requires motion processing. However, here we ask whether, relative to within pathway integration, across pathway integration may incur additional processing costs because color is processed in the ventral stream, while with motion is processed in the dorsal stream.

Feature integration relies on efficient connectivity between visual processing regions (e.g., Festa et al., 2005). While feature integration within a visual pathway likely relies on short, local connections within each region of the visual hierarchy, feature integration across visual pathways, in addition, likely relies on more distant, distributed connections between visual processing regions. Coincidentally, connectivity exhibits dynamic changes, from short- to long-range, across child development (Cao, Huang, & He, 2017; Fair et al., 2007, 2009; Supekar, Musen, & Menon, 2009; Uddin, Supekar, & Menon, 2010), providing a unique opportunity to examine distinct feature integration across the visual cortical hierarchy. Together this suggests that, earlier in childhood, integrating features processed in separate pathways (e.g., color and motion) may come with an additional processing cost relative to integrating features processed within the same visual pathway (e.g., luminance and motion). Put another way, the additional processing cost of integrating features across visual pathways may decrease across childhood. Within the same child, an additional cost for color-motion integration, relative to luminance-motion integration, should differentially impact conjunction visual search performance depending on the visual features that define the targets and distractors. This result would strongly suggest that developing visual function is an agent of change in VSA development (Amso & Scerif, 2015). In order to isolate the change in color-motion feature integration relative to global improvements in information processing, we examine color-motion feature integration performance in relation to luminance-motion feature integration.

In the current study, 4- to 10-year-old children performed a visual search task. In two conditions, children were asked to search for a moving target that varied by either color or by luminance, placing more or less demand on feature integration across the visual pathways. In both Feature conditions, targets are presented with no distractors, or with 2 or 4 distractors. In the absence of distractors, performance reflects baseline feature integration abilities. In the presence of distractors, performance reflects visual selective attention abilities; specifically, the change in performance with an increase in distractor number (i.e., search slope). We first predicted that color-motion integration would be associated with a performance cost, relative to luminance-motion integration, and that this cost would decrease with age as color-motion integration improves from early to middle childhood. We next predicted that, across early to middle childhood, VSA for color-motion would change more than VSA for luminance-motion. Finally, we predicted that individual differences in developing feature integration may be associated with developmental changes in VSA. Specifically, developmental improvements in feature integration should be associated with steeper visual search slopes, and this should be more evident for the color-motion than luminance-motion visual search conditions. As children become better at integrating color and motion, sensitivity to the conjunction of features that define competing distractors should increase. This would result in a greater amount of time needed to resolve visual competition during target selection.

Methods

Participants

Eighty-nine 4- to 10-year-old children (Overall: M = 7.17, SD = 1.82, Range = 4.14 −10.75, 39 female; Female: M = 7.44, SD = 1.99, Range = 4.14 – 10.75; Male: M = 6.96, SD = 1.66, Range = 4.18 – 10.26) comprised the final sample. Children were normally distributed across age (Skewness Z = 0.62). An additional 12 children were tested, but excluded due to non-compliance (n = 4), experimenter or technical error (n = 3), or color blindness (n = 5). We removed 5 children as multivariate outliers and 4 children as univariate outliers, and 10 children that did not contribute data for selective attention trials (e.g., no correct Set Size 3 or 5 trials, see below). Children and their parents were recruited through advertisements and were all local community members. Children provided assent and adults provided consent in accordance with the University IRB. Families were compensated 15 US dollars for their time.

Children’s race make-up included 78% White, 7% Multi-racial, 8% Black/African-American, 6% “Other”, and 2% declined to answer. Ethnicity make-up included 84% non-Hispanic, 14% Hispanic, and 2% declined to answer. Participants’ average IQ, as determined by the Woodcock-Johnson Brief Intelligence Assessment (Woodcock, McGrew, & Mather, 2007), was M= 109.53, SD= 16.08 points. One child did not complete IQ testing.

Stimuli & Apparatus

Stimuli consisted of red, green, white, and black circles (approximately 1.25° in diameter) that moved either vertically or horizontally in phase synchrony. Circles oscillated approximately 1.25° in either direction around their initial starting point at a speed of approximately 3° s−1. Using a ColorCal MKII colorimeter (Cambridge Research Systems), we measured the luminance (Y) and Commission Internationale de l’Eclairage (CIE) coordinates (x, y) of the stimuli. Luminance-matched red (Y = 19 cd/m2; x = 0.60; y = 0.34) and green circles (Y = 19 cd/m2; x = 0.32, y = 0.51) appeared on a black background (Y = 0.25 cd/m2; x = 0.26; y = 0.26). Chromaticity-matched black (Y = 0.25 cd/m2; x = 0.26; y = 0.26) and white circles (Y = 185 cd/m2; x = 0.33; y = 0.32) appeared on a gray background (Y = 16.10 cd/m2; x = 0.314; y = 0.342). Circles were presented in one of six concentric locations equidistant from the screen center (approximately 6°), where an orange cartoon clown fish (“Nemo”) served as a fixation point. Children were allowed to move their eyes freely throughout the trials. Within a given trial, children saw a search display (Figure 1) for up to 3500ms. If a response was recorded, the search display was removed. Following each search display, a cartoon fish was presented for 1000ms, to direct children’s attention to the center of the screen.

Figure 1.

Figure 1.

Open in a new tab

Illustrations of search displays for Feature and Set Size conditions for both Target-Present and Target-Absent trials. The left-most columns depict target-present and target-absent color-motion integration trials. The right-most columns depict target-present and target-absent luminance-motion integration trials. Rows depict Set Size trials, within an increase in distractors from top to bottom. The top row depicts Feature Integration trials. The bottom two rows depict Visual Selective Attention trials. Target stimuli are highlighted by a dotted yellow circle. White arrows were not presented to participants, but instead represent motion direction. Distractors could differ from the arget in either color or luminance, but share vertical motion. Or, distractors could differ in motion direction, but share either color or luminance value. Each display was presented until the child responded (up to 3500 ms) and was followed by a fixation display (1000 ms).

Procedure

Children were first screened for color blindness using the Ishihara tests for color-deficiency. All children in the final sample passed these tests and showed no evidence of color blindness. Prior to the trials of interest, children were also asked to point to, or verbally discriminate between, red and green, and black and white circles, as well as vertical (or ‘jumping’) and horizontal (or ‘sideways’) motion. Children were instructed to “press the button as quickly as you can”, once they found the target on target-present trials, and were instructed to “not press the button” on target-absent trials. Next, children completed two practice trials to ensure they understood the instructions. This procedure was repeated if children failed to correctly indicate a color or motion direction, or if they incorrectly responded to either practice trial. Children were then asked to verbally indicate the target stimulus identity (“a jumping red/black circle”) to the experimenter. Children then searched for a vertically moving target circle among distractor circles. Across two Feature conditions, we manipulated which visual feature required integration with motion. In the luminance-motion Feature condition, the target was a vertically moving black circle, and distractors were vertically moving white circles and horizontally moving black circles. In the color-motion Feature condition, the target was a vertically moving red circle, and distractors were vertically moving green circles and horizontally moving red circles. Thus, children were required to integrate motion with either luminance or color information. We also manipulated the number of stimuli presented within each Feature condition. Across three Set Size conditions, stimuli were presented in sets of 1, 3 or 5. Target circles were present in 50% of trials and absent in 50% of trials. Target locations was randomly selected. Feature conditions (luminance-motion and color-motion) were blocked and counterbalanced. Set Size conditions (1, 3, 5) were pseudorandomly ordered. In total, children completed 96 trials, 48 for each Feature condition (luminance-motion and color-motion) and 16 (8 target-present, 8 target-absent) for each Set Size within each Feature condition. Each child was offered a short break every 24 trials. Figure 1 illustrates sample search displays for each Feature, Set Size and Trial Type (target-present, target-absent).

Dependent Measures

For each Feature and Set Size condition, we recorded RTs on target-present trials and calculated target detection sensitivity (d’) across target-present and target-absent trials. Initial data inspections revealed that accuracy was at ceiling in many cases, across many conditions. Thus, we applied a log-linear correction to the calculation of d’ (Hautus, 1995; Stanislaw & Todorov, 1999). Briefly, .5 was added to both Hit Rates and False Alarm Rates and 1 was added to both the number of target present and target absent trials. We then calculated d’ by subtracting the normalized False Alarm Rate from the normalized Hit Rate.

Feature Integration Performance.

We define feature integration as the detection sensitivity for a target defined by multiple visual features (e.g., Treisman, 1998), without spatially competing distractors. Children were instructed to press a button when they found the target stimulus. Targets were either present (e.g., vertically moving red circle) or absent (e.g., horizontally moving red circle or vertically moving green circle). We generated a target detection sensitivity (d’) value for each Feature condition, when targets were presented without distractors (Set Size 1).

We also created a Feature Integration Index to measure the added cost of integrating color and motion features relative to luminance and motion features. To do this, we subtracted each participant’s luminance-motion from color-motion integration performance value. A larger negative Feature Integration Index reflects greater performance cost for color-motion relative to luminance-motion feature integration, while a positive reflects greater performance cost for luminance-motion relative to color-motion feature integration. A value of zero thus reflects no performance cost for either color-motion and luminance-motion feature integration. Since this index was significantly skewed (Z = −2.95), we rank this measure to reduce skewness (Z = −0.65).

Visual Selective Attention Performance.

We measure visual selective attention (VSA) performance as the change in children’s performance as a function of distractor number (i.e., search slope). We thus calculated the performance slope for both reaction time (RT) and target detection sensitivity (d’) as the ratio of change in performance across Set Size over the change in Set Size. We then control for age-related differences in manual dexterity across our wide age range by dividing this performance slope value by performance on Set Size 1 trials. This estimates the visual search rate (e.g., RT slope) proportional to each individual child’s baseline performance. Thus, larger RT search slope values reflect slower visual search rates, while smaller RT search slope values reflect faster visual search rates. In contrast, smaller d’ search slope values reflect greater influence of distractors on accuracy, while larger d’ search slope values reflect smaller influence of distractors on accuracy.

Results

Feature Integration Performance

Following the removal of outliers, additional outliers were revealed and Feature Integration measures remained skewed (color-motion d’: Z = −7.79, luminance-motion d’: Z = −8.69, color-motion RT: Z = 3.90, luminance-motion RT: Z = 5.55). To reduce the potential influence of outliers and skewness we first collapsed across Feature conditions and then rank-transformed each Feature Integration measure, resulting in less skewed distributions (color-motion d’: Z = −2.39, luminance-motion d’: Z = −3.54, color-motion RT: Z = −0.11, luminance-motion RT: Z = 0.10).

We predicted that color-motion Feature Integration would be associated with a performance cost, relative to luminance-motion integration, and that this cost would decrease with age as color-motion integration improves from early to middle childhood. To test this prediction, we submitted both Feature Integration performance measures (Set Size 1 ranked RT for correct target-present trials only and Set Size 1 ranked target detection sensitivity) to separate repeated measures ANCOVAs with Feature condition (color-motion, luminance-motion) as a within-subjects variable and Age (in years) as a continuous variable. See Table 1 for dependent variable descriptive statistics, collapsed across age.

Table 1.

Summary of Behavioral Performance Measures

Color-motion Luminance-motion
Feature Integration Visual Selective Attention Feature Integration Visual Selective Attention
d′ RT d′ RT d′ RT d′ RT
N 89 89 87 89 89 89 89 89
Mean 2.70 1141.90 −0.04 0.11 2.95 1152.75 −0.05 0.12
Median 3.19 1080.13 0.00 0.10 3.19 1084.00 −0.06 0.10
Standard 0.79 279.14 0.15 0.13 0.45 311.21 0.11 0.13
Deviation
Minimum −0.31 692.75 −0.50 −0.25 1.00 722.00 −0.43 −0.21
Maximum 3.19 2123.83 0.36 0.46 3.19 2475.00 0.16 0.46
Open in a new tab

For correct target-present RTs, we only found a main effect of Age, F(1,87)=57.073, p < .001, partial eta = .396, all other p’s > .865. We thus submitted the unranked (raw) mean RTs, collapsed across Feature conditions, to Spearman’s ranked correlations and found that that RTs decreased with age rs(89) = −.608, p < .001. This indicates that children become faster to correctly detect a target across childhood.

For d’, there was a main effect of Feature condition, F(1,87) = 9.454, p = .003, partial eta = .098, where color-motion integration was worse than luminance-motion integration (Table 1). There was also a main effect of Age, F(1,87) = 27.112, p < .000, partial eta = .238, where overall target detection sensitivity improved across early to middle childhood, rs(89) = .564, p < .001. As predicted, there was additionally an Age by Feature condition interaction, F(1,87) = 6.362, p = .013, partial eta = .068. To understand the interaction, we submitted raw (unranked) d’ measures to Spearman’s ranked correlations and found that, while both Feature conditions show age-related improvement, Age was correlated with color-motion integration, rs(89) = .529, p < .001, to a greater extent than luminance-motion integration, rs(89) = .255, p = .016.

Figure 2 shows that color-motion integration is worse than luminance-motion integration in early childhood, but feature integration becomes equivalent by middle childhood. This finding is consistent with our prediction that color-motion feature integration would be associated with a performance cost relative to luminance-motion integration. However, the added cost of binding color and motion across visual pathways decreases across childhood. In other words, color-motion integration, relative to luminace-motion integraiton improves across childhood.

Figure 2.

Figure 2.

Open in a new tab

Age-related changes in Feature Integration accuracy, as measured by target detection sensitivity (d’) for Set Size 1 trials. Color-motion target detection sensitivity increased with Age to a greater extent than luminance-motion target detection sensitivity.

Visual Selective Attention Performance

Age-related changes.

We predicted that, across childhood, visual search performance for color-motion targets would change more than search for luminance-motion target. To test this prediction, we submitted baseline-corrected search slopes for each dependent variable (RT and d’) in separate repeated measures ANCOVAs with Feature condition (luminance-motion, color-motion) as a within subject variable and Age (in years) as a continuous variable. We found no effects for the baseline-corrected d’ search slopes, all p’s > .06.

For baseline-corrected RT search slopes, we found a main effect of Feature condition, F(1,87) = 5.236, p = .025, partial eta = .057, where search rates were slower for luminance-motion relative to color-motion conditions. There was also a main effect of Age, F(1,87) = 13.315, p < .001, partial eta = .133. Pearson correlations showed that, an increase in distractor number was associated with greater slowing for search with Age, r(89) = .364, p < .001. Critically, there was also an Age by Feature condition interaction, F(1,87) = 4.917, p = .029, partial eta = .053, suggesting luminance-motion and color-motion visual search change with Age differently across childhood. Figure 3A shows that color-motion, r(89) = .432, p < .001, but not luminance-motion, r(89) = .125, p = .242, visual search performance changed across early to middle childhood. These data suggest, as predicted, that luminance-motion visual search is stable earlier than color-motion. Moreover, the pattern of results shows that children have steeper color-motion RT slopes with Age, indicating that they become more sensitive to additional distractors with Age in the color-motion search condition only.

Figure 3.

Figure 3.

Open in a new tab

A) Age-related changes in Visual Selective Attention Performance. Color-motion visual search rates slowed across childhood, but this effect is not evidence in luminance-motion visual search. B) Individual differences in Feature Integration predict Color-motion Visual Selective Attention Performance. Worse color-motion integration, relative to luminance-motion integration, is associated with slower color-motion visual search rates. Raw Feature Binding Index is plotted for easier interpretation.

Individual differences in Feature Integration.

We predicted that individual differences in Feature Integration may influence VSA across childhood. In particular, stronger visual feature integration should strengthen both target and distractor processing and therefore increase competition with increasing distractor number. This would result in slower visual search as distractor number increases.

To test this prediction, we submitted baseline-corrected search slopes for each dependent variable (RT and d’) to separate repeated measures ANCOVAs with Feature condition (luminance-motion, color-motion) as a within subject variable and the ranked Feature Integration Index (i.e., difference score between Set Size 1 d’ color-motion – luminance-motion) as a continuous variable. See Table 1 for descriptive statistics for each condition. For baseline-corrected d’ search slopes, we found no significant effects, all p’s > .169.

For RTs, we found a main effect of Feature condition, F(1, 87) = 5.779, p = .018, partial eta = .062, with slower search rates for the luminance-motion condition relative to the color-motion condition. There was no main effect of Feature Integration Index, p = .294. However, there was a Feature condition by Feature Integration Index interaction, F(1, 87) = 5.800, p = .018, partial eta = .062. Figure 3B shows that when luminance-motion Feature Integration is better than the color-motion, luminance-motion RT visual search slopes are steeper. Within the same child, as color-motion Feature Integration performance approached luminance-motion performance, this difference in visual search RT slope values decreased. Thus, as color-motion Feature Integration performance approached luminance-motion performance, visual search RT slopes increased for the color-motion, rs(89) = .243, p = .022, but not for luminance-motion condition, rs(89) = −.067, p = .533 (Figure 3B). Thus, improvements in color-motion Feature Integration, relative to luminance-motion, resulted in greater slowing for color-motion search.

So far, we have shown that (1) Feature Integration performance for color-motion targets improves with Age (Figure 2), (2) VSA performance on color-motion trials reflects increased sensitivity to distractors with Age (Figure 3A), and (3) improvement in color-motion, relative to luminance-motion, Feature Integration is associated with greater sensitivity to color-motion distractors during visual search (Figure 3B). These findings indicate that the relationship between Age, Feature Integration, and VSA in our age-range is specific to the color-motion visual search condition. To directly test this claim, we submitted baseline-corrected RT search slopes to a repeated measures ANCOVA with Feature condition (luminance-motion, color-motion) as a within subject factor and the Age by Feature Integration Index interaction as a covariate (or continuous variable). As before, we found a main effect of Feature condition, F(1,86) = 7.882, p = .006, partial eta = .083, where search rates were slower for luminance-motion search relative to color-motion search. We also found a Feature condition by Age by Feature Integration interaction, F(1,86) = 8.475, p = .005, partial eta = .089. These results suggest that, regardless of a child’s age, when luminance-motion integration is better than color motion integration, luminance-motion search rates are slower overall. In contrast, as color-motion feature integration comes to approximate luminance-motion feature integration with age, color-motion search rates become slower. When feature integration is equal, however, both search rates are similar and color-motion search rates slow with age. Thus, age-related changes in color-motion integration may increase children’s sensitivity to color-motion distractors, as revealed by steeper visual search RT slopes.

Discussion

We examined whether children’s feature integration and visual selective attention abilities for objects in motion change with age. First, we found that, while feature integration improved with age, this effect was larger for color-motion integration relative to luminance-motion integration. This suggests that while color-motion integration was worse than luminance-motion integration in early childhood, the two become equivalent by middle childhood. Second, while RT search slopes were, on average, steeper for the luminance-motion condition, slopes increased with age for the color-motion condition. This result revealed that while luminance-motion search performance was robust across childhood, older children were more influenced by additional color-motion distractors. Third, when luminance-motion integration was better than color-motion integration, luminance-motion RT search slopes were steeper, indicating that children were more sensitive to the addition of luminance-motion distractors that competed with the to-be selected target. In contrast, age-related improvements in color-motion feature integration were associated with steeper RT search slopes.

Our results add to the visual search developmental literature in two important ways. First, we demonstrate that, in the absence of distractors, younger children are worse at integrating multiple visual features relative to older children. This pattern was especially evident for color-motion integration relative to luminance-motion integration. Prior work found that, in the absence of distractors, both children and adults were slower at detecting a target defined by two features compared to a target defined by one feature (Trick & Enns, 1998). Our results are consistent with the interpretation that younger children, relative to older children, are slower to integrate feature information during conjunction visual search. We add that integrating feature information across parallel visual streams may be costlier early in childhood relative to integrating feature information within a visual stream.

Second, the present study examined how differences in feature integration impact visual selective attention. To our knowledge, ours is the first study to examine, conjunction visual search performance as a function of variable feature integration demands across or within visual pathways within the children. Previous work in adults has shown that visual search performance varies by visual sensitivity (Hunter, Godde, & Olk, 2018; Li, Sampson, & Vidyasagar, 2007). Previous work in children has shown that distractor number (Donnelly et al., 2007; Gerhardstein & Rovee-Collier, 2002) and top-down cues (Lookadoo, Yang, & Merrill, 2017; Merrill & Lookadoo, 2004) impact developmental visual search trajectories. Still other work has shown that basic oculomotor information processing and improvement in visuospatial abilities impact visual search development across adolescence (Burggraaf, van der Geest, Hooge, & Frens, 2019). Our work is also consistent with recent work showing that the ability to track a moving target among distractors improves across late childhood (Wolf et al., 2018). We found that, relative to the color-motion visual search, children were slowed by increasing distractor set size more when searching for luminance-motion targets, but this effect was constant across the 4–10-year-old age-range. In contrast, color-motion visual search became slower with additional distractors with age, and this slowing was associated with age-related improvements in color-motion feature integration. Visual search performance, thus, depends on many factors, which may differentially influence this ability at different times in development. Moreover, together these findings show that there is no single visual search developmental trajectory, but multiple developmental trajectories that likely interact across development. Future work will consider whether feature integration across and within visual pathways is stable by adolescence, and if so whether visual search slopes would then show a decline with age, perhaps reflecting general information processing mechanisms.

These data suggest that the development of the attentional mechanisms that support learning and memory (Markant & Amso, 2014; Markant, Worden, & Amso, 2015; Werchan, Lynn, & Kirkham, 2019) may be impacted by the changes in robustness of visual processing across childhood (Amso & Scerif, 2015). The present study provides evidence that, across early to middle childhood, as the ability to integrate color and motion visual features improves, competition between targets and distractors may increase, thereby increasing the time needed to resolve this competition by processing additional distractors during the target selection process. Thus, developmental changes in visual feature integration abilities may be important for developmental changes in VSA. These findings have important implications for developmental work showing that learning and memory for features processed in separate visual pathways may follow distinct developmental trajectories (Lange-Küttner & Küttner, 2015), which may be related to visual processing development (see, Braddick & Atkinson, 2011). Indeed, visual acuity, luminance and chromatic contrast sensitivity (e.g., Bradley, Arthur and Freeman, 1982; Ellemberg, Lewis, Hong Liu, & Maurer, 1999; Knoblauch, Vital-Durand, & Barbur, 2001), and global motion direction sensitivity (e.g., Ellemberg, Lewis, Maurer, Brar, & Brent, 2002; Hadad, Maurer, & Lewis, 2011) all improve across childhood. Moreover, some suggest that luminance thresholds necessary for form perception improve from middle to late childhood (Bertone, Hanck, Guy, & Cornish, 2010). Future work will examine the impact of visual feature processing development on feature integration abilities across early to middle childhood.

Our findings also mirror those from patients with Alzheimer’s disease (AD), whose cortical connectivity is disrupted (see, Delbeuck, Linden, & Collette, 2003). AD patients exhibit greater age-related slowing for conjunction visual search when compared to healthy elderly adults (Foster, Behrmann, & Stuss, 1999). AD patients are better at detecting global motion that requires feature integration within one visual pathway relative to feature integration between distinct, parallel visual pathways (Festa et al., 2005). Thus, greater improvement in color-motion feature integration across childhood, relative to luminance-motion feature integration, may suggest that integration across relatively distinct visual pathways may develop later in childhood than integration features processed within a single visual pathway. This age-related improvement in pathway integration is in line with developmental patterns of network connectivity (Cao et al., 2017; Fair et al., 2007, 2009; Hagmann et al., 2010; Supekar et al., 2009; Uddin et al., 2010) and increasing coherence across visual cortices during childhood (Kipping, Tuan, Fortier, & Qiu, 2017). Future work will also examine whether feature integration reflects underlying functional connectivity within and between visual pathways.

Conclusions

The current study adds to the developmental literature by showing that visual systems development is an agent of change in VSA development (Amso, Haas, & Markant, 2014; Amso & Scerif, 2015). Feature integration within a visual pathway may develop earlier than between visual pathways and thus differentially impact target-distractor similarity during the visual selective attention process across childhood. Indeed, as color-motion integration improved across childhood, visual search rates slowed, suggesting competition between color-motion defined targets and distractors increased. Mechanistically, as distractor number increases, robust color-motion feature integration across the visual scene would mean more locations competing for selection. Children with relatively better between visual pathway integration may, therefore, be more sensitive to color-motion distractors relative to luminance-motion distractors and may need additional time to recognize distractors as non-targets during the selection process. These data suggest that VSA development may be better conceptualized as a biased competition computation (e.g., Desimone & Duncan, 1995) rather than a finite discernible network of attentional processes with a uniform developmental trajectory (Petersen & Posner, 2012).

Research highlights.

  • Feature integration improves across childhood

  • In early childhood, color-motion is worse than luminance-motion integration

  • Luminance-motion visual search rate does not change across childhood

  • Color-motion visual search rate slows across childhood

  • Color-motion integration improvements predict slower color-motion search rates

Acknowledgements:

We thank the members of the Developmental Cognitive Neuroscience Lab at Brown University (especially Diego Placido) for help with recruitment and data collection, and all of the children and families who made this research possible. This work was funded by an NSF Graduate Research Fellowship (to AL), the James S. McDonnell Scholar Award in Understanding Human Cognition (to DA), and by NIH R21-MH113870 (to DA), and by R01 MH099078 (to DA).

Footnotes

Conflict of Interest Statement: The authors have no conflicts of interest to declare.

References

  1. Amso D, Haas S, & Markant J (2014). An eye tracking investigation of developmental change in bottom-up attention orienting to faces in cluttered natural scenes. PLoS ONE, 9(1), 1–7. 10.1371/journal.pone.0085701 [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Amso D, & Scerif G (2015). The attentive brain: insights from developmental cognitive neuroscience. Nature Reviews Neuroscience, 16(10), 606–619. 10.1038/nrn4025 [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Bertone A, Hanck J, Guy J, & Cornish K (2010). The development of luminance-and texture-defined form perception during the school-aged years. Neuropsychologia, 48(10), 3080–3085. 10.1016/j.neuropsychologia.2010.06.019 [DOI] [PubMed] [Google Scholar]
  4. Braddick O, & Atkinson J (2011). Development of human visual function. Vision Research, 51(13), 1588–1609. 10.1016/j.visres.2011.02.018 [DOI] [PubMed] [Google Scholar]
  5. Bradley, Arthur and Freeman RD (1982). Contrast sensitivity in children. Vision Research, 22, 953–959. http://doi.org/0262121794,9780262121798 [DOI] [PubMed] [Google Scholar]
  6. Burggraaf R, van der Geest JN, Hooge ITC, & Frens MA (2019). Developmental changes in visual search are determined by changing visuospatial abilities and task repetition: a longitudinal study in adolescents. Applied Neuropsychology: Child, 0(0), 1–11. 10.1080/21622965.2019.1627211 [DOI] [PubMed] [Google Scholar]
  7. Cao M, Huang H, & He Y (2017). Developmental Connectomics from Infancy through Early Childhood. Trends in Neurosciences, 40(8), 494–506. 10.1016/j.tins.2017.06.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Delbeuck X, Linden M Van Der, & Collette F (2003). Alzheimer’s Disease as a Disconnection Syndrome ?, 13(2). [DOI] [PubMed] [Google Scholar]
  9. Desimone R, & Duncan J (1995). Neural Mechanisms of Selective Visual Attention. Annual Review of Neuroscience, 18(1), 193–222. 10.1146/annurev.ne.18.030195.001205 [DOI] [PubMed] [Google Scholar]
  10. Donnelly N, Cave K, Greenway R, Hadwin JA, Stevenson J, & Sonuga-Barke E (2007). Visual search in children and adults: Top-down and bottom-up mechanisms. Quarterly Journal of Experimental Psychology, 60(1), 120–136. 10.1080/17470210600625362 [DOI] [PubMed] [Google Scholar]
  11. Ellemberg D, Lewis TL, Hong Liu C, & Maurer D (1999). Development of spatial and temporal vision during childhood. Vision Research, 39(14), 2325–2333. 10.1016/S0042-6989(98)00280-6 [DOI] [PubMed] [Google Scholar]
  12. Ellemberg D, Lewis TL, Maurer D, Brar S, & Brent HP (2002). Better perception of global motion after monocular than after binocular deprivation. Vision Research, 42(2), 169–179. 10.1016/S0042-6989(01)00278-4 [DOI] [PubMed] [Google Scholar]
  13. Fair DA, Cohen AL, Power JD, Dosenbach NUF, Church JA, Miezin FM, … Petersen SE (2009). Functional Brain Networks Develop from a ‘“Local to Distributed”‘ Organization. PLoS Comput Biol, 5(5), 14–23. 10.1371/journal.pcbi.1000381 [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Fair DA, Dosenbach NUF, Church JA, Cohen AL, Brahmbhatt S, Miezin FM, … Schlaggar BL (2007). Development of distinct control networks through segregation and integration. Proceedings of the National Academy of Sciences of the United States of America, 104(33), 13507–13512. 10.1073/pnas.0705843104 [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Felleman DJ, & Van Essen DC (1991). Distributed hierarchical processing in the primate cerebral cortex. Cerebral Cortex, 1, 1–47. [DOI] [PubMed] [Google Scholar]
  16. Festa EK, Insler RZ, Salmon DP, Paxton J, Hamilton JM, & Heindel WC (2005). Neocortical disconnectivity disrupts sensory integration in Alzheimer’s disease. Neuropsychology, 19(6), 728–738. 10.1037/0894-4105.19.6.728 [DOI] [PubMed] [Google Scholar]
  17. Foster JK, Behrmann M, & Stuss DT (1999). Visual attention deficits in Alzheimer’s Disease: Simple versus conjoined feature search. Neuropsychology, 13(2), 223–245. [DOI] [PubMed] [Google Scholar]
  18. Gegenfurtner KR (2003). Cortical mechanisms of colour vision. Nature Reviews Neuroscience, 4(7), 563–572. 10.1038/nrn1138 [DOI] [PubMed] [Google Scholar]
  19. Gerhardstein P, & Rovee-Collier C (2002). The development of visual search in infants and very young children. Journal of Experimental Child Psychology, 81(2), 194–215. 10.1006/jecp.2001.2649 [DOI] [PubMed] [Google Scholar]
  20. Hadad BS, Maurer D, & Lewis TL (2011). Long trajectory for the development of sensitivity to global and biological motion. Developmental Science, 14(6), 1330–1339. 10.1111/j.1467-7687.2011.01078.x [DOI] [PubMed] [Google Scholar]
  21. Hagmann P, Sporns O, Madan N, Cammoun L, Pienaar R, Wedeen VJ, … Grant PE (2010). White matter maturation reshapes structural connectivity in the late developing human brain. Proceedings of the National Academy of Sciences, 107(44), 19067–19072. 10.1073/pnas.1009073107 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Hautus MJ (1995). Corrections for extreme proportions and their biasing effects on estimated values of d’. Behavior Research Methods, Instruments & Computers, 27(1), 46–51. [Google Scholar]
  23. Hommel B, Li KZH, & Li S-C (2004). Visual Search Across the Life Span. Developmental Psychology, 40(4), 545–558. 10.1037/0012-1649.40.4.545 [DOI] [PubMed] [Google Scholar]
  24. Hunter M, Godde B, & Olk B (2018). Effects of absolute luminance and luminance contrast on visual search in low mesopic environments. Attention, Perception, and Psychophysics, 80(5), 1265–1277. 10.3758/s13414-018-1512-0 [DOI] [PubMed] [Google Scholar]
  25. Kipping JA, Tuan TA, Fortier MV, & Qiu A (2017). Asynchronous Development of Cerebellar, Cerebello-Cortical, and Cortico-Cortical Functional Networks in Infancy, Childhood, and Adulthood. Cerebral Cortex, 27(11), 5170–5184. 10.1093/cercor/bhw298 [DOI] [PubMed] [Google Scholar]
  26. Knoblauch K, Vital-Durand F, & Barbur JL (2001). Variation of chromatic sensitivity across the life span. Vision Research, 41(1), 23–36. 10.1016/S0042-6989(00)00205-4 [DOI] [PubMed] [Google Scholar]
  27. Lange-Küttner C, & Küttner E (2015). How to learn places without spatial concepts: Does the what-and-where reaction time system in children regulate learning during stimulus repetition? Brain and Cognition, 97, 59–73. 10.1016/j.bandc.2015.04.008 [DOI] [PubMed] [Google Scholar]
  28. Li JCH, Sampson GP, & Vidyasagar TR (2007). Interactions between luminance and colour channels in visual search and their relationship to parallel neural channels in vision. Experimental Brain Research, 176(3), 510–518. 10.1007/s00221-006-0773-3 [DOI] [PubMed] [Google Scholar]
  29. Lobaugh NJ, Cole S, & Rovet JF (1998). Visual search for features and conjunctions in development. Canadian Journal of Experimental Psychology, 52(4), 201–211. [DOI] [PubMed] [Google Scholar]
  30. Lookadoo R, Yang Y, & Merrill EC (2017). Encouraging top-down attention in visual search:A developmental perspective. Attention, Perception, and Psychophysics, 79(7), 2007–2020. 10.3758/s13414-017-1379-5 [DOI] [PubMed] [Google Scholar]
  31. Markant J, Ackerman LK, Nussenbaum K, & Amso D (2016). Selective attention neutralizes the adverse effects of low socioeconomic status on memory in 9-month-old infants. Developmental Cognitive Neuroscience, 18, 26–33. 10.1016/j.dcn.2015.10.009 [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Markant J, & Amso D (2013). Selective Memories: Infants’ Encoding is Enhanced in Selection Via Suppression. Developmental Science, 16(6), 926–940. 10.1111/desc.12084.Selective [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Markant J, & Amso D (2014). Leveling the playing field: attention mitigates the effects of intelligence on memory. Cognition, 131(2), 195–204. 10.1016/j.cognition.2014.01.006 [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Markant J, Worden MS, & Amso D (2015). Not all attention orienting is created equal: Recognition memory is enhanced when attention orienting involves distractor suppression. Neurobiology of Learning and Memory, 120, 28–40. 10.1016/j.nlm.2015.02.006 [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Merrill EC, & Lookadoo R (2004). Selective search for conjunctively defined targets by children and young adults. Journal of Experimental Child Psychology, 89(1), 72–90. 10.1016/j.jecp.2004.04.001 [DOI] [PubMed] [Google Scholar]
  36. Petersen SE, & Posner MI (2012). The Attention System of the Human Brain: 20 Years After. Annual Review of Neuroscience, 35(1), 73–89. 10.1146/annurev-neuro-062111-150525 [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Seymour K, Clifford CWG, Logothetis NK, & Bartels A (2009). The Coding of Color, Motion, and Their Conjunction in the Human Visual Cortex. Current Biology, 19(3), 177–183. 10.1016/j.cub.2008.12.050 [DOI] [PubMed] [Google Scholar]
  38. Shipp S, & Zeki S (1995). Segregation and convergence of specialised pathways in macaque monkey visual cortex. Journal of Anatomy, 187 (Pt 3, 547–62. 10.21037/atm.2016.10.49 [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Sincich LC, & Horton JC (2005). The Circuitry of V1 and V2: Integration of Color, Form, and Motion. Annual Review of Neuroscience, 28(1), 303–326. 10.1146/annurev.neuro.28.061604.135731 [DOI] [PubMed] [Google Scholar]
  40. Stanislaw H, & Todorov N (1999). Calculation of signal detection theory measures. Behavior Research Methods, Instruments, & Computers, 31(1), 137–149. 10.3758/BF03207704 [DOI] [PubMed] [Google Scholar]
  41. Supekar K, Musen M, & Menon V (2009). Development of large-scale functional brain networks in children. PLoS Biol, 7(7), e1000157. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Treisman A (1998). Feature binding, attention and object perception. Philosophical Transactions of the Royal Society B: Biological Sciences, 353(1373), 1295–1306. 10.1098/rstb.1998.0284 [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Treisman A, & Gelade G (1980). A feature-integration theory of attention. Cognitive Psychology, 12, 97–136. [DOI] [PubMed] [Google Scholar]
  44. Trick LM, & Enns JT (1998). Lifespan changes in attention: the visual search task. Cog Dev, 13, 369–386. [Google Scholar]
  45. Uddin LQ, Supekar K, & Menon V (2010). Typical and atypical development of functional human brain networks: insights from resting-state fMRI. Frontiers in Systems Neuroscience, 4(May), 1–12. 10.3389/fnsys.2010.00021 [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Ungerleider LG, & Haxby JV (1994). “What” and “where” in the human brain. Current Opinion in Neurobiology, 4(2), 157–165. 10.1016/0959-4388(94)90066-3 [DOI] [PubMed] [Google Scholar]
  47. Werchan DM, Lynn A, & Kirkham NZ (2019). The emergence of object - based visual attention in infancy: A role for family socioeconomic status and competing visual features. Infancy, (Advance online publication), 1–16. 10.1111/infa.12309 [DOI] [PubMed] [Google Scholar]
  48. Wolf K, Weber EG, Bosch JJF Van Den, Volz S, Nöth U, Deichmann R, … Fiebach CJ (2018). Neurocognitive Development of the Resolution of Selective Visuo-Spatial Attention: Functional MRI Evidence From Object Tracking, 9(July), 1–16. 10.3389/fpsyg.2018.01106 [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Wolfe JM (1994). Guided search 2.0 a Revised Model of Visual Search, 1(2), 202–238. 10.3758/BF03200774 [DOI] [PubMed] [Google Scholar]
  50. Woodcock RW, McGrew KS, & Mather N (2007). Woodcock-Johnson III. Rolling Meadows, IL: Riverside Publishing. [Google Scholar]
  51. Zeki SM (1978). Functional specialisation in the visual cortex of the rhesus monkey. Nature, 274(3). [DOI] [PubMed] [Google Scholar]

RESOURCES