حواس پرتی در آشکارسازی ADHD در مورد مکانیسم برای کنترل توجه بالا به پایین؟

Abstract In this study, we attempted to clarify whether distractibility in ADHD might arise from increased sensory-driven interference or from inefficient top-down control. We employed an attentional filtering paradigm in which discrimination difficulty and distractor salience (amount of image “graying”) were parametrically manipulated. Increased discrimination difficulty should add to the load of top-down processes, whereas increased distractor salience should produce stronger sensory interference. We found an unexpected interaction of discrimination difficulty and distractor salience. For difficult discriminations, ADHD children filtered distractors as efficiently as healthy children and adults; as expected, all three groups were slower to respond with high vs. low salience distractors. In contrast, for easy discriminations, robust between-group differences emerged: ADHD children were much slower and made more errors than either healthy children or adults. For easy discriminations, healthy children and adults filtered out high salience distractors as easily as low salience distractors, but ADHD children were slower to respond on trials with low salience distractors than they did on trials with high salience distractors. These initial results from a small sample of ADHD children have implications for models of attentional control, and ways in which it can malfunction. The fact that ADHD children exhibited efficient attentional filtering when task demands were high, but showed deficient and atypical distractor filtering under low task demands suggests that attention deficits in ADHD may stem from a failure to efficiently engage top-down control rather than an inability to implement filtering in sensory processing regions.

. Introduction Attention deficit hyperactivity disorder (ADHD) is the most common childhood mental disorder, affecting between 5% and 10% of children worldwide (Faraone, Sergeant, Gillberg, & Biederman, 2003). Hyperactivity, impulsivity, and inattention are all major behavioral symptoms of ADHD. However, while many studies have documented that ADHD children are impaired in executive functions, including response inhibition, working memory, and conflict resolution (Bush et al., 1999, Casey et al., 1997, Doyle, 2006, Pliszka et al., 2006, Rubia et al., 2005, Schulz et al., 2004 and Vaidya et al., 2005), the nature and extent of attention deficits in ADHD remain controversial. Although ADHD children are typically slower and more variable to respond to cued targets (Nigg et al., 1997, Novak et al., 1995 and van der Meere and Sergeant, 1988), ADHD children have not previously been reported to be impaired at filtering out irrelevant distractors. Healthy and ADHD children exhibit similar patterns of slowed responses to relevant targets when distractors are present (Booth et al., 2005, Huang-Pollock and Nigg, 2003, Huang-Pollock et al., 2005, Huang-Pollock et al., 2006, Mason et al., 2003, Mason et al., 2005, Nigg et al., 1997, Novak et al., 1995, Oberlin et al., 2005 and van der Meere and Sergeant, 1988). This has led several researchers to question whether selective attention is a core deficit in ADHD or whether attentional problems are secondary to deficits of alertness (Huang-Pollock et al., 2005) or other executive processes, including inhibition (Barkley, 1997). The current study aimed to better characterize the nature of attention deficits in ADHD, viewed from the context of the biased competition model of attention (Desimone and Duncan, 1995, Kastner and Ungerleider, 2000 and Kastner and Ungerleider, 2001). According to this model, limited neural and cognitive resources necessitate privileged processing of some sensory inputs and associated responses at the expense of others. Limited capacity of cortical sensory regions leads to bottom-up, perceptual interference from competing stimuli (Desimone and Duncan, 1995, Kastner et al., 1998, Moran and Desimone, 1985 and Reynolds et al., 1999) such that distractors reduce the magnitude and efficiency of neural and behavioral responses. However, stimulus-driven sensory competition can be overcome by top-down, intentional feedback from a network of prefrontal and parietal regions (Kastner et al., 1999, Kastner and Ungerleider, 2000 and Kastner and Ungerleider, 2001). But while prefrontal and parietal cortex can mediate sensory competition in visual regions, these top-down sources have their own capacity limits. For example, performance on tasks that draw heavily on executive functions – such as tasks with high working memory load – can deteriorate due to insufficient prefrontal capacity to support efficient attentional filtering. (Lavie & DeFockert, 2003; Lavie, 2005). In the current study, we hoped to gain insight into the functional locus of attention deficits in ADHD. Specifically, is distractibility caused by increased competition in sensory cortex, decreased capacity of cognitive control regions, or deficient feedback from control areas to sensory regions? Furthermore, if ADHD can be shown to selectively impair bottom-up or top-down processes, then our findings would provide evidence for modularity and independence of sensory competition and top-down attentional control. To isolate sensory-level and top-down components of distractor filtering, distractor salience and task difficulty were both parametrically manipulated in an orthogonal fashion. To probe sensory interactions, we manipulated perceptual load by varying distractor salience. Increasing distractor salience has been shown to diminish perceptual responses to target stimuli in ventral stream visual areas (Desimone and Duncan, 1995, Moran and Desimone, 1985 and Reynolds et al., 1999). To test the integrity or efficiency of top-down control regions, we manipulated discrimination difficulty in a face discrimination task. We believe that task difficulty was a measure of cognitive load because (1) the face discrimination task involved a comparison of a presented face to an iconic image; (2) the judgment was based on slight differences between morphed face images rather than low-level visual features such as oriented edges or shape; and (3) perceptual decision-making has been shown to be mediated by regions of frontal cortex (Heekeren, Marrett, Bandettini, & Ungerleider, 2004). Healthy children (age 8–13), ADHD children (age 8–13), and healthy adults practiced a face discrimination task and their perceptual threshold was measured in a staircase procedure. This allowed us to tailor task difficulty to each individual’s perceptual threshold. What behavioral patterns were expected for ADHD children? First, if distractibility in ADHD children results from deficient filtering mechanisms in sensory areas, we would expect to see distractor-dependent behavioral deficits. These would manifest as greater interference from high salience distractors than from low salience distractors in ADHD, compared to controls. An inability to filter out the suppressive effects of distractors in sensory areas should produce steeper RT x distractor salience slopes in ADHD than healthy subjects, similar to the effects of lesions of extrastriate visual processing areas V4 and TEO ( Buffalo et al., 2005, De Weerd et al., 1999 and Gallant et al., 2000). We would not expect sensory-driven filtering deficits to be influenced by discrimination difficulty. Alternatively, if distractibility in ADHD results from decreased prefrontal and parietal capacity for top-down modulation, we would expect a different pattern of results. Specifically, more challenging tasks should create more competition for limited resources, and, in turn, greater decrements in distractor filtering in ADHD relative to healthy children. Similarly, if distractibility is due to diminished strength of top-down control, then high salience distractors, which cause the largest sensory interference, would require the strongest top-down control. Thus performance in ADHD relative to healthy children would be most affected by high salience distractors, especially for resource–intensive difficult discriminations. Finally, if distractibility is not due to diminished capacity or strength of top-down signals, but instead reflects a heightened threshold for recruiting top-down control, then ADHD children should be more distractible when deployment of selective attention is under endogenous control and not task-driven. In this instance, we expected that ADHD children would be most distractible when performing easy compared to hard discriminations. The current study focuses primarily on differences between healthy and ADHD children. However, neurocognitive deficits in ADHD have been attributed to neurodevelopmental immaturity. Thus, for two reasons, inclusion of data from healthy adults also clarifies the nature of any detected performance differences between ADHD and healthy children. First, because selective attention has been studied more extensively with adults than children, most theories of attention are based on data from adults. Inclusion of healthy adults in the current study provides a benchmark against which healthy and ADHD children can be compared, facilitating integration of current theories focused narrowly on attention and on ADHD. Second, data in healthy adults clarifies potential developmental influences on task performance, which in turn shapes views of ADHD as arising from neurodevelopmental immaturity (Shaw et al., 2007).

Results During perceptual training, we found that healthy adults had significantly lower perceptual thresholds than healthy children (t = 2.87, p = .004) and ADHD children (t = 1.92, p = .033), but ADHD and healthy children did not differ. Thus, when searching for targets, adults discriminated more subtle differences among morphed images than healthy and ADHD children, who required bigger differences between the stimuli in order to discriminate between targets and non-targets. Turning next to the experimental blocks with randomly varying distractor salience, we found a significant main effect of task difficulty (response times, F(1,41) = 129.5, p < .001; error rates, F(1,41) = 231.9, p < .001). Participants took longer to respond (802 ms ± 20) and made more errors (28% ± 1) for difficult relative to easy discriminations (600 ms ± 15; 3% ± 0). We also found a significant main effect of subject group (response times, F(2,41) = 4.47, p = .017; error rates, F(2,41) = 8.0, p = .001). ADHD children responded significantly slower (809 ms ± 20) than healthy children (631 ms ± 26, t = 5.43, p < .00001) or adults (661 ms ± 21, t = 5.08, p < .00001). There was no significant difference in response time between healthy children and adults. Both ADHD and healthy children made more errors (ADHD: 17% ± 2, t = 2.76, p = .003; healthy children: 19% ± 2, t = 3.48, p = .0003) than healthy adults (11% + 1) but did not differ from each other. There was no main effect of distractor salience. There was a significant interaction of discrimination difficulty with subject group (response times, F(2,41) = 6.88, p = .003; error rates, F(2,41) = 4.5, p = .017). This interaction reflected the fact that the largest differences between ADHD children and other subject groups emerged during easy, compared to difficult, discriminations. For difficult discriminations ( Fig. 2a), ADHD children responded slower (884 ms ± 31) than healthy adults (806 ms ± 24, t = 2.01, p = .024) and children (709 ms ± 41, t = 3.40, p < .001), whereas healthy children responded significantly faster than healthy adults (t = −2.01, p = .024). For difficult discriminations ( Fig. 3a), ADHD children made more errors (28% ± 2) than healthy adults (21% ± 2, t = 2.60, p < .001) but fewer errors than healthy children (35% ± 1, t = −2.86, p < .005). The difference in error rate between healthy children and healthy adults was also significant (t = 6.49, p < .00001). For easy discriminations ( Fig. 2c), ADHD children were much slower to respond (734 ms ± 22) than healthy adults (515 ms ± 15; t = 8.16, p < .000001) and healthy children (553 ± 25; t = 5.37, p < .000001). For easy discriminations, there were no significant differences in response time between healthy children and adults. For easy discriminations ( Fig. 3b), ADHD children made significantly more errors (5% ± 1) than healthy adults (0%, t = 4.58, p < .0001) and children (3% ± 0; t = 2.18, p = .016). The difference between healthy children and adults was also significant (t = 5.41, p < .0001). As noted above, these between-group differences for easy discriminations were larger than those found for difficult discriminations, as reflected in the group-by-difficulty interactions. In a and c, response time is plotted as a function of distractor salience for ... Fig. 2. In a and c, response time is plotted as a function of distractor salience for difficult and easy discriminations, for each of the three participant groups. In b and d, filtering cost (the intra-individual difference in response time for trials with high salience distractors and response time for trials with low salience distractors) is plotted for difficult and easy discriminations, for each subject group. Figure options Error rate as a function of distractor salience for easy and difficult ... Fig. 3. Error rate as a function of distractor salience for easy and difficult discriminations. Figure options There was also a significant interaction of task difficulty with distractor salience (response times, F(2,82) = 3.71, p = .029; error rates, n.s.) This interaction reflects the fact that response times increased as a function of distractor salience for difficult disciminations (low salience distractors: 793 ms ± 35; medium salience distractors: 802 ms ± 33; high salience distractors: 811 ms ± 34), but there was no linear increase in response time for easy disciminations (low salience distractors: 607 ms ± 26; medium salience distractors: 601 ms ± 26; high salience distractors: 597 ms ± 25). The difference in response time between difficult discriminations with low and high salience distractors was significant (t = −1.95, p = .029); no other contrasts reached significance. Inspection of Fig. 2a and c, suggests that the subject groups showed different patterns of interaction of distractor salience and task difficulty. Nonetheless, the triple interaction of difficulty × salience × subject group did not reach significance. However, the main effect of task difficulty indicated that there were very large differences in response times between easy and difficult discriminations. This large main effect of task difficulty and inter-subject variability may have obscured smaller differences in response times between trials with low and high salience distractors. Consequently, in further analysis, we focused on the within-subject “filtering costs”. For difficult and easy discriminations, we calculated the “filtering cost” imposed by distractors of increasing salience, as described in the methods. Discrimination difficulty and subject group did not have significant main effects on filtering cost, but there was a significant interaction of the two factors (F(2,39) = 3.82, p < .05). For difficult discriminations, all three groups showed the same pattern of results with increasing response times as a function of distractor salience ( Fig. 2a) and positive filtering costs ( Fig. 2b). There was no significant difference between-groups in the magnitude of filtering costs for difficult discriminations (healthy adults: 18 ± 14 ms; healthy children: 20 ± 15 ms; ADHD children: 17 ± 20 ms, all p’s > .40). However, groups manifested distinct performance patterns for easy discriminations ( Fig. 2d). Healthy adults and children did not pay a significant filtering cost during easy discriminations: Their filtering costs (adults: 10 ± 5 ms; children: −2 ± 12 ms) did not differ from each other (t = −0.88, p = .20) and were not significantly greater than zero (adults: t = 1.7, p > .10; children: t = −.1, p > .10). Unexpectedly, ADHD children had a negative filtering cost (−37 ± 17 ms) which was significantly less than zero (t = −2.15, p < .05), indicating that they were slower to respond on trials with low salience distractors than on trials with high salience distractors. For easy discriminations, the filtering costs of ADHD children were significantly different than those of healthy children (t = −1.69, p = .05) and adults (t = −2.58 p < .01), as shown in Fig. 2d. We also looked for differences in response variability between ADHD children and healthy peers, especially because intra-individual variability has been noted as a very reliable behavioral marker of ADHD (Castellanos and Tannock, 2002 and Castellanos et al., 2005). Table 1 lists the coefficient of variation (CV) for all three subject groups as a function of discrimination difficulty and distractor salience. There was a main effect of group (F = 6.49, p < .005), with healthy adults exhibiting lower intra-individual variability (mean CV, collapsed across experimental factors: .255) than healthy children (mean CV: .370, t = 3.09, p < .005) and ADHD children (mean CV: .398, t = 5.15, p < .0001). Although there was a trend toward ADHD children showing more variability than healthy children, there was no statistical difference in average CV for the two groups (t = 0.65, p = .26). Intra-individual variability was not correlated with task difficulty or distractor salience (i.e. no main effects), nor did these experimental factors significantly interact with the main effect of group, although there was a trend toward an interaction of group and task difficulty (F = 1.86, p = .17), with increased variability in ADHD children relative to healthy children when performing easy discriminations. Table 1. Intra-individual variability: mean (SEM) coefficient of variation. Group Difficult discrimination, low salience distractors Difficult discrimination, medium salience distractors Difficult discrimination, high salience distractors Easy discrimination, low salience distractors Easy discrimination, medium salience distractors Easy discrimination, high salience distractors Healthy adults .259 (.015) .269 (.012) .251 (.012) .243 (.031) .238 (.022) .247 (.019) Healthy children .362 (.034) .378 (.026) .388 (.042) .332 (.036) .359 (.041) .383 (.045) ADHD children .378 (.029) .376 (.028) .390 (.033) .404 (.026) .419 (.030) .396 (.031) Table options Because we found significant differences in error rates between-groups—namely that ADHD children made more errors than healthy children and adults for easy discriminations (Fig. 3b) and that healthy children made more errors than adults and ADHD children for difficult discriminations (Fig. 3a)—we explored whether response time differences were due to speed-accuracy trade-offs. Error rates were inversely correlated with response time (Fig. 4b) for healthy children (r2 = 0.41), but not for healthy adults (r2 = .17; Fig. 4a), suggesting that healthy children but not adults employed a strategy with a speed-accuracy trade-off. Unlike healthy children, ADHD children did not exhibit a speed-accuracy trade-off (r2 = .01; Fig. 4c). Hence, the increased error rate of ADHD children for easy discriminations reflects greater distractor interference on target processing rather than a change in motor response strategies (i.e. more impulsive button pressing). Analysis of speed-accuracy trade-off for healthy adults, healthy children, and ... Fig. 4. Analysis of speed-accuracy trade-off for healthy adults, healthy children, and ADHD children.