Specific Differences in Temporal Binding Aspects of the Attentional Blink in Chromosome 22q11.2 Deletion Syndrome

Abstract

Chromosome 22q11.2 Deletion Syndrome (22q11DS) is a genetic syndrome characterized by a variety of cognitive impairments, including difficulty with attention. 22q11DS is the strongest known genetic risk factor for developing schizophrenia, a disorder characterized by impairments in visual attention and temporal binding processes. Here we examine a specific temporal visual attention phenomenon (the attentional blink; AB) within two rapid serial visual presentation tasks, and compare those with 22q11DS to groups of typically developing individuals matched on chronological (CA) and mental age (MA). Performance of individuals with 22q11DS was sensitive to differing task demands. On a Category Task, individuals with 22q11DS performed similarly to control groups on all measures of the AB, with the exception of lower detection accuracy of the first of two targets. In contrast, on a feature-based Color Task which required temporal binding of stimulus features, individuals with 22q11DS differed from CA and MA matched control groups on all AB performance measures, exhibiting lower target accuracy, more temporal binding errors, and a deeper, more protracted AB. Temporal binding in the visual domain is thought to be dependent on a serial attention mechanism that facilitates simultaneous firing of neurons in multiple areas of the visual cortex, activating short-term working memory for storage of bound features. Given the discrepancy between these two tasks, results suggest that temporal binding processes may be significantly affected in individuals with 22q11DS, a finding that importantly, has been previously demonstrated among individuals with schizophrenia.^*

1. Introduction

Chromosome 22q11DS11.2 Deletion Syndrome (22q11DS), previously known as Velo-Cardio-Facial Syndrome (VCFS), is a genetic syndrome caused primarily by de novo microdeletion of approximately 3 million base pairs of DNA encompassing over 40 protein-coding genes on the long (q) arm of chromosome 22 (Dunham, Shimizu, Roe & Chissoe, 1999; Morrow et al., 1995). 22q11DS is currently the second most common genetic syndrome, after Down syndrome, affecting approximately 1 in every 4000 to 6000 births (Botto et al., 2003; Driscoll et al., 1993; Gothelf & Lombroso, 2001). Although its phenotype can vary across individuals, the syndrome is often characterized by specific physical features (Gothelf et al., 1999; Shprintzen et al., 1978), developmental and cognitive delays (average IQ’s are in the low 70s but have been found to range from 50-109; De Smedt et al., 2007; Óskarsdóttir et al., 2005), as well as a variety of behavioral and psychiatric problems (Antshel et al., 2007; Stevens & Murphy, 2005). Prevalent psychological co-morbid disorders include attention deficit hyperactivity disorder (ADHD), autism spectrum disorder, generalized anxiety disorder, and perhaps most uniquely, schizophrenia (Antshel et al., 2006; Antshel et al., 2007; Feinstein, Eliez, Blasey & Reiss, 2002; Green et al., 2009). In fact, 22q11DS is currently the strongest know molecular genetic risk factor for developing schizophrenia (McDonald-McGinn et al., 2016) with occurrence rates as high as 20-30% (Murphy, Jones & Owen, 1999; Pulver et al., 1994; Shprintzen, Goldberg, Golding-Kushner & Marion, 1992).

Studies of individuals with 22q11DS have focused on identifying familial, genetic, structural and functional correlates and interactions that confer increased risk for schizophrenia among those with 22q11DS, including studies that attempted to differentiate those with 22q11DS who do, and do not, go on to develop schizophrenia (Antshel et al., 2010; Schneider et al., 2014). These studies have provided important insights into risk and resilience factors, as well as the potential common mechanisms underlying 22q11DS and schizophrenia syndromes. However, fewer experimental studies directly examined the cognitive, attentional and perceptual phenotypes of 22q11DS, which could help inform our understanding of specific strengths and weaknesses that characterize the disorder, more generally, and how this phenotype relates to schizophrenia more specifically.

The association of attentional difficulties among populations of individuals with schizophrenia is ubiquitous (Luck & Gold, 2008), beginning with Bleuler’s original descriptions of Dementia Praecox (1911). Though attention is a broad construct, findings in the schizophrenia literature indicate that individuals with schizophrenia have a difficult time with particular aspects of attention related to attentional control. More specifically, data converge on the notion that individuals with schizophrenia have difficulty with the processes that manage where and when to allocate attention, especially in situations that require rapid and efficient visual processing. However, once they have ‘locked on’ to a target, they are not impaired in the processes required to identify it (Fuller et al., 2006; Luck, & Gold, 2008). In contrast to this precise characterization in schizophrenia, there is a paucity of research examining attentional processing in 22q11DS despite clinical descriptions of attentional difficulties (Woodin et al., 2001), as well as well-established associations with both schizophrenia and other disorders related to attention such as ADHD.

Experimental studies that have investigated visual attention among persons with 22q11DS suggest that perhaps the same characteristic performance patterns noted in schizophrenia might also be present in 22q11DS. For example, participants did not differ from peers on simple tasks with few demands on attentional selection, such as those that require the detection of a single stimulus in the absence of distractors, (Simon et al., 2008) or on valid trials of a spatial cueing paradigm (Simon, Bearden, McGinn, & Zackai, 2005). In contrast, impairments emerged on tasks that made demands on attentional selection, such as when participants were required to inhibit their attention to an invalid cue (Bish, Chiodo, Mattei, & Simon, 2007; Simon et al., 2005) or invalid distractors (Bish, Ferrante, McDonald-McGinn, Zackai, & Simon, 2005; Sobin, Kiley-Brabeck, Daniels & Blundell, 2004) that were presented temporally synchronous, or asynchronous with a target.

The aforementioned studies provide information relevant to the understanding of attention processing in individuals with 22q11DS. However, they have often looked concurrently at spatial and temporal attentional functioning within the same paradigm, and critically, the performance of individuals with 22q11DS was always compared to that of typically developing (TD) participants matched on chronological age, despite the presence of intellectual delays in this population. Here we address these two issues by comparing the performance of participants with 22q11DS with two groups of TD individuals, one matched on chronological age, and one matched on mental age, which is an alternative to matching by intelligence quotient (IQ; Burack et al., 2012; Burack, Iarocci, Bowler & Mottron, 2002; Burack, Iarocci, Flanagan & Bowler, 2004; Hodapp & Dykens, 2001) on two tasks that make different demands on visual attention in the temporal domain using a well-established phenomenon known as the attentional blink. The use of a within-subjects design across two tasks with built-in control measures, as well as the use of developmentally appropriate sample matching strengthen our experiment and improve our ability to accurately identifying specific group differences (Chapman & Chapman, 1973; Strauss, 2001).

1.1. Measuring the Deployment of Attention and Temporal Integration of Episodic Information via the Attentional Blink

The attentional blink (AB) is a phenomenon that has been useful in understanding temporal attention processing in the visual domain. The AB is a metaphorical blink of attention that is reflected in a decrease in the detection accuracy of the second of two targets presented in a spatially fixed, sequential order, when it occurs within 200-500ms of a first target (Forster, 1970; Potter, 1984). In a typical trial, target and distractor stimuli are presented rapidly in sequence at the same location on a display at rates of 10 items per second (Forster, 1970; Potter, 1984). Participants are told which target(s) to watch for in the stream, and are asked to report them at the end of each trial. In these paradigms, stimuli can be symbols, digits, letters, words, or pictures and targets often differ from distractor stimuli by feature (e.g. color) or by category (e.g. letters or numbers). The task nuances require different strategies and therefore often produce slightly differential AB patterns.

1.2. The Attentional Blink in Schizophrenia

There are currently six existing studies investigating the AB in schizophrenia (Cheung, Chen, Chen, Woo & Yee, 2002; Li et al., 2002; Mathis, Wynn, Breitmeyer, Nuechterlein & Green, 2011; Mathis et al., 2012; Su et al., 2015; Wynn, Breitmeyer, Nuechterlein & Green, 2006) and all of these have used categorical designs (participants had to identify target letters in a stream of distractor numbers). In all but one (Su et al., 2015), individuals with schizophrenia showed significantly poorer target detection accuracy compared to TD individuals matched by chronological age. All five of these studies reported an exaggerated or deeper AB in those with schizophrenia and four of the studies (Cheung et al., 2002; Li et al., 2002; Mathis et al., 2011; Mathis et al., 2012) reported a protracted AB effect in schizophrenia patients compared to age-matched controls, indicating attentional impairment beyond the typical AB window of 200-500ms. Further, the AB was reported to be exaggerated in individuals with schizophrenia relative to healthy comparison participants, even when controlling for overall poorer performance on a single-target RSVP task (Cheung et al., 2002; Mathis et al., 2011).

In contrast, a recent study investigating the AB among individuals with schizophrenia (Su et al., 2015) took a different approach, attempting to control for attentional impairments that are known to be present in individuals with schizophrenia. They achieved this by slowing down the stimulus presentation rate for those with schizophrenia (1 per 100ms) compared to control participants (1 per 50ms), so that single-target accuracy was similar between the two groups. Su et al. (2015) found that, with this control in place, individuals with schizophrenia performed similarly to TD individuals, suggesting that differences in the AB in schizophrenia may in fact be accounted for by developmental differences or delays, rather than impairments specific to the disorder.

In addition to categorical tasks, feature-based AB tasks allow for the measurement of temporal binding of stimulus features (e.g. binding the target color to the target stimulus) and the proportion of binding errors made. It has been proposed that temporal binding in the visual domain is mediated by the simultaneous (on the timescale of several milliseconds) firing of neurons in multiple areas of the visual cortex, each specialized to process distinct features of objects (e.g., color, form, etc.; Crick & Koch, 1990; Engel, Fries, König, Brecht, & Singer, 1999). Substantial evidence for this synchronous firing has been documented in higher primates (Engel, Roelfsema, Fries, Brecht, & Singer, 1997), particularly in the visual domain, though this process is likely to extend to other sensory modalities which are also topographically organized into specialized processing areas in the cortex (Engel et al., 1999). It has also been suggested that important sensory regions of the brain, such as the thalamus and basal ganglia, may play an important role in this process (Crick & Koch, 1990). Conscious awareness of visual sensations is thought to rely on some form of serial attention mechanism that facilitates the simultaneous firing of sets of relevant neurons, allowing for the activation of short-term working memory where bound features of an object may be temporarily stored (Crick & Koch, 1990). Given that abnormal cortical organization has been documented in individuals with 22q11DS (Schmitt et al., 2015), processes that may require specific integration between cortical areas, including temporal binding, may be impacted in this population.

The used of feature-based tasks may also be important to investigate in individuals with 22q11DS given the previous literature suggesting temporal binding difficulties in individuals with schizophrenia (Martin, Giersch, Huron, & van Wassenhove, 2013; Parsons et al., 2013; Stevenson et al., 2017). For example, Martin et al. (2013) tested patients with schizophrenia and age-matched comparison participants on the McGurk task, in which mouth movements and auditory signals representing two different syllables are fused together to create a third, illusory syllable (McGurk & MacDonald, 1976). The authors also varied the stimulus onset asynchronies (SOA) between the auditory and visual stimuli. Among those who succumbed to the illusion, individuals with schizophrenia reported fusion over much longer SOA’s than the comparison participants, suggesting that binding is extended beyond the typical temporal window of integration in this population. This finding of an extended binding window has also been corroborated in simultaneity judgement tasks, in which participants were asked to determine whether auditory and visual stimuli that consisted of speech and non-speech (in separate tasks) were presented simultaneously or not (Stevenson et al., 2017), and suggest that visual temporal attention might be particularly important to examine among persons with 22q11DS.

1.3. Predictions of the Current Study

We hypothesized that individuals with 22q11DS would show similar performance on measures of the AB in comparison to mental age (MA) matched TD control participants, but impaired performance relative to chronological age-matched peers (CA) on a categorical task, in which we required participants to detect two letters in a stream of numbers. However, given findings of temporal binding difficulties among those with schizophrenia, we expected that on a color-based task, in which participants were asked to detect two purple letters in a string of black letters, the performance of individuals with 22q11DS would be impaired relative to both matched groups (CA and MA) as reflected in reduced detection accuracy of the first (T1) and second (T2) targets, slower recovery from the AB, and a larger proportion of swaps (reporting both targets correctly but in the incorrect order) and binding errors (erroneously bind the color to distractor items before or after the target).

We also explored the relationship between performance on the AB tasks and behavioral markers of prodromal schizophrenia symptoms using the Scale of Prodromal Symptoms (SOPS; Miller et al., 2003).

2. Method

2.1. Participants

Participants with 22q11DS were recruited through the Center for the Diagnosis, Treatment, and Study of Velo-Cardio-Facial Syndrome (VCFS) at SUNY-Upstate Medical University from a larger sample of participants in a longitudinal psychosis risk factors study. All participants had a fluorescence in situ hybridization (FISH)-confirmed deletion in the q11.2 region of chromosome 22. Participants were excluded if they had any other identifiable neurological condition besides 22q11DS that may affect cognitive performance. Eligible and willing participants took part in this experiment as part of their fourth visit for the longitudinal study.

Typically developing (TD) children were recruited through word of mouth and through flyers placed in the community and delivered via school listservs. TD adults were recruited through word of mouth and through the SONA research system, in which undergraduate students in Psychology participate in research to earn course credit. Participants were excluded if they reported non-corrected vision problems. TD child participants were also excluded if their parents reported the child as having a history of academic or psychiatric problems.

Sixteen individuals with 22q11DS initially participated in the study. One participant’s data was excluded due to poor performance, with accuracy on the task greater than 2 standard deviations below the mean, indicating that the participant either did not understand, or did not follow the directions. This resulted in a total of 15 participants with 22q11DS included in the analyses. The Wechsler Adult Intelligence Scale, Third Edition (WAIS-III; Wechsler, 1997) was used to measure the intelligence quotient (IQ) and subsequently calculate the mental age of each 22q11DS participant. Each participant with 22q1 IDS was then matched to TD participants by chronological age (CA) and by mental age (MA), resulting in two TD control groups (see Table 1 for participant characteristics and medication information). Since IQ measures were not administered to TD participants, MA was not calculated for these groups and was assumed to be approximately equal to CA, as would be expected in typically developing individuals.

Table 1.

Participant Characteristics and Medication Information

Group	N	CA in years (SD)	MA in years (SD)	N on Medication
22q11DS	15 (7 female)	20.74 (2.5)	17.02 (3.5)	6a
CA matched	15 (9 female)	20.86 (2.36)	-	0
MA matched	15 (12 female)	17.06 (3.71)	-	0

Notes. CA = chronological age, MA = mental age, SE = standard error of measurement

^aIndicates the number of participants that were taking one or more medications (stimulants, benzodiazepines, anti-depressants or anti-anxiety) to manage symptoms associated with 22q11DS during their participation in the study.

2.2. Experimental Task

All participants completed two rapid serial visual presentation (RSVP) attentional blink tasks: a Category Task and a Color Task. The order in which these tasks were completed was randomly counterbalanced across participants in each group.

In the Category Task, participants were required to identify targets that differed from distractors by category; i.e. targets were letters and distractors were numbers. Two black letters (the targets) were embedded in a stream of black number distractors, and participants were asked to identify the two black letters at the end of each RSVP stream. Target stimuli included black uppercase letters A, B, C, D, F, H, J, and K, while distractor stimuli included black numbers 1, 2, 3, 4, 5, 6, 7, 8, and 9.

In the Color Task, two purple letters (the targets) were embedded in a stream of black letter distractors, and participants were asked to identify the purple letters at the end of each RSVP stream. Target stimuli included purple uppercase letters A, B, C, D, F, H, J, K, L, N, P, R, T, V, X, and Y, while distractor stimuli included the same uppercase letters in black text.

Participants completed 234 trials of each task and were allowed to take a short break between tasks if needed. Stimuli in the RSVP stream were presented at a rate of 1 item per 135ms with no inter-stimulus-interval. All stimuli were presented in 48 point Arial font in the center of the screen on a light grey background. Each trial consisted of 30 stimuli; 2 targets and 28 distractors. Trials began with a fixation cross in the center of the screen, presented for 200ms, followed by the RSVP stream, illustrated in Figure 1. The first target (T1) was presented anywhere between position 6 and 18 in the stream. The second target (T2) was presented anywhere between one and eight lags after T1. Lag indicates what position in the RSVP stream T2 is presented following T1 (i.e. lag 1 indicates that T2 is presented directly after T1 in stream, whereas lag 8 indicates that T2 is presented 8 positions after T1 with 7 distractors in between). This resulted in the following target onset asynchronies: Lag 1 (135ms), lag 2 (270ms), lag 3 (405ms), lag4 (540ms), lag 5 (675ms), lag 6 (810ms), lag 7 (945ms), and lag 8 (1080ms). As such, the lag 2 and lag 3 presentations of T2 would fall within the 200-500ms attentional blink period. T1 position and T1-T2 lag were counterbalanced to ensure that these were evenly mixed.

Figure 1.

RSVP paradigm and response screens for the Category (left) and Color (right) attentional blink tasks. In the Color task, target letters are purple rather than white as shown here.

Following each trial, two identical response screens were presented (one at a time) for the identification of T1 and T2. Each screen contained two rows of possible targets, presented on a light grey background, along with the statements ‘I saw nothing’ and ‘I don’t know what I saw’ underneath. All participants were asked to verbally report which uppercase letter targets they saw in order, while the experimenter entered their responses on a standard keyboard. See Figure 1 for a visual representation of the RSVP paradigm used in both the Color and Category Tasks.

All stimuli were presented using Matlab on either a Dell P2210 with a resolution of 1680 by 1000 pixels with a 60 Hz refresh rate or a Macintosh Mini with a screen resolution of 1680 by 1050 pixels with a 60 Hz refresh rate. Despite the slight difference in screen resolution, the visual angle of all stimuli was kept constant, subtending 1.2 degrees vertically and 1.3 degrees horizontally from the center of the screen. The programs were initiated and programmed using Stream, a Matlab toolbox that uses Psychophysics toolbox (Brainard, 1997).

2.3. Procedure

Informed consent was obtained from all participants 18 years and older, while parent or guardian consent and participant assent were obtained for individuals under the age of 18 years. All procedures were approved by the IRB of the respective institutions where data were collected. To assess for behavioral markers of schizophrenia, individuals with 22q11DS were assessed for positive prodromal symptoms of schizophrenia using the Positive Symptom Subscale of the Scale of Prodromal Symptoms (SOPS). IQ and SOPS data were collected at the same time point in the longitudinal study as this experiment took place (the fourth visit). All participants then completed the AB tasks. Following completion of the study, participants were compensated for their participation, either monetarily ($10 per hour of participation) or with SONA course credit.

3. Analysis Strategy & Results

To investigate differences in the accuracy of target reporting, T1 accuracy and T2|T1 accuracy were calculated for each participant at each lag. T2|T1 represents the proportion of trials for each lag in which participants correctly identified T2 given that T1 was also correctly identified on that trial. In this study, T2 and T1 were considered correct even in instances when their temporal order was swapped (i.e. when T2 was reported as T1 and vice versa; Russo et al., 2016). Proportion of swaps was measured as the number of trials in which T1 and T2 were correctly identified but in reverse order, compared to the total number of correct T1 and T2 trials. An omnibus repeated measures (RM) ANOVA on T1 Accuracy revealed a significant main effect of task type (category vs. color; F_GG^* (1, 42) = 17.21, p < .001, η² = .29, 95% CI [.08; .47])), so the tasks were analyzed independently. For each task, separate RM ANOVAs were conducted for T1 accuracy, T2|T1 accuracy, and proportion of swaps, with the between-subjects factor of group (22q11DS, MA-matched controls, CA-matched controls) and the within-subjects factor of lag (8).

Depth of the AB was measured by subtracting the mean accuracy at lags 2 and 3 (the AB period) from the mean at lags 7 and 8 (see Colzato, Spape, Pannebakker, & Hommel, 2007; Kelly & Dux, 2011; Martens & Johnson, 2009; Russo et al., 2016) and rate of recovery from the AB was calculated for each participant in each task by determining the lag (between lags 3-8) at which T2|T1 accuracy reached 80% of maximum accuracy (based on highest accuracy score at lags 4-8 (see Dye & Bavelier, 2010; Russo, Kates, & Wyble, 2017). One-way ANOVAs were conducted to evaluate differences between groups for both depth of the AB and rate of recovery from the AB.

In the Color Task, binding errors were assessed by calculating the proportion of trials in which participants inaccurately reported non-target letters at adjacent temporal positions (+/− 5, 4, 3, 2 and 1) to the actual target. An omnibus RM ANOVA revealed a main effect of target (T1 vs. T2; F_GG(1, 402) = 12.39, p < .001, η² = .03, 95% CI [.006; .07]), therefore separate RM ANOVAs were conducted for each target.

3.1. Category Task

3.1.1. T1 accuracy

(see Supplemental Figure 1): There was a main effect of group (F(2, 42) = 9.14, p = .001, η² = .3, 95% CI [.08; .47]) in which the 22q11DS group (M_22q11DS = .81, SE = .021) showed significantly lower T1 accuracy than both MA-matched controls (M_MA = .90, SE = .021, p = .005) and CA-matched controls (M_CA = .94, SE = .021, p < .001), who performed similarly (p = .24). There was also a main effect of lag (F_GG(5.46, 229.13) = 3.42, p = .004, η² = .08, 95% CI [.009; .13]) in which overall T1 Accuracy at lag 1 (M_lag1 = .84, SE = .019) was significantly lower than accuracy at all other lags (ps <.04) and accuracy at lag 2 (M_lag2 = .88, SE = .016) was lower than accuracy at lag 7 (M_lag7 = .90 , SE = .014, p = .04). There were no interactions between group and lag, suggesting that patterns of T1 accuracy were similar across groups.

3.1.2. T2|T1 accuracy

(see Figure 2): There was a main effect of lag (F_GG(3.1, 130.3) = 30.54, p < .001, η² = .42, 95% CI [.28; .51]) in which accuracy at lags 2, 3, and 4 were significantly lower than accuracy at all other lags (ps < .001), suggesting a typical AB pattern. T2|T1 accuracy at lag 5 was also lower than accuracy at lags 7 and 8 (ps <.03), indicating an overall recovery period from the AB. Neither a main effect of group nor an interaction between lag and group were observed, suggesting that there were no differences in the level or pattern of performance between groups in terms of T2|T1 accuracy.

Figure 2.

T2|T1 accuracy by lag in the Category attentional blink task. Error bars represent +/− 1 standard error.

3.1.3. Proportion of swaps

(see Supplemental Figure 2): There was a main effect of lag (F_GG(2.86, 120.19) = 38.4, p < .001, η² = .48, 95% CI [.34; .58]) in which a significantly higher proportion of swaps were made at lag 1 compared to all subsequent lags (ps < .001), as well as at lag 2 compared to all following lags (ps ≤ .02). Neither a main effect of group nor an interaction between lag and group were observed, suggesting that there were no differences in the proportion or pattern of swaps between groups.

3.1.4. Depth of the AB and rate of recovery.

There were no significant differences in the depth of the AB between groups (F(2, 42) = .031, p = .97), suggesting that all groups experienced similar presence of and recovery from the attentional blink in the Category Task. Further, there were no significant differences in the rate of recovery from the AB between groups (F(2, 42) = .014, p = .99).

3.2. Color Task

3.2.1. T1 accuracy

(see Supplemental Figure 3): There was a main effect of group (F(2,42) = 14.89, p < .001, η² = .41, 95% CI [.17; .56]) in which the 22q11DS group (M_22q11DS = .72, SE = .025) showed significantly lower T1 accuracy than both MA-matched controls (M_MA = .87, SE = .025, p = .001) and CA-matched controls (M_CA = .91, SE = .025, p < .001), who performed similarly (p = .31). There was also a main effect of lag (F(7, 294) = 6.54, p < .001, η² = .13, 95% CI [.05; .19]), but no interactions between group and lag, suggesting that patterns of T1 accuracy were similar across groups despite overall lower accuracy in the 22q11DS group relative to the two control groups.

3.2.2. T2|T1 accuracy

(see Figure 3): There was a main of group (F(2, 42) = 18.3, p < .001 , η² = .47, 95% CI [.22; .61]) in which those with 22q11DS (M_22q11DS = .62, SE = .028) showed significantly lower accuracy than both the MA-matched (M_MA = .82, SE = .028, p < .001) and CA-matched (M_CA = .85, SE = .028, p < .001) control groups, who performed similarly (p = .46). There was also a main effect of lag (F_GG(4.19, 175.91) = 38.76, p < .001, η² = .48, 95% CI [.37; .55]) that followed a typical AB pattern, with lower accuracy at lags 2 and 3, and a slow recovery of performance. However, the key finding here is an interaction between group and lag (F_GG(8.37, 175.91) = 4.92, p < .001, η² = .19, 95% CI [.06; .26]), in which the 22q11DS group showed significantly lower T2|T1 accuracy at lags 2 through 7 than both control groups (ps < .02), which did not differ significantly from each other. These results suggest that individuals with 22q11DS may experience specific impairment in temporal visual attention processing under conditions which require temporal binding (e.g., binding the color purple to a specific letter in a stream of rapidly presented letters), as evidenced by a deeper AB and a slower recovery, which is further examined below.

Figure 3.

T2|T1 accuracy by lag in the Color attentional blink task. Error bars represent +/− 1 standard error.

3.2.3. Proportion of swaps

(see Figure 4): There was a main effect of group (F(2, 42) = 10.64, p < .001, n² = .34, 95% CI [.10; .50]), in which the 22q11DS group (M_22q11DS = .043, SE = .006) showed significantly more swaps than both the MA- matched (M_MA = .017, SE = .006, p = .002) and CA-matched (M_CA = .008, SE = .006, p < .001) control groups, which did not differ from each other (p = .3). There was also a main effect of lag (F_GG(1.84, 77.34) = 30.98, p < .001, η² = .42, 95% CI [.25; .54]), in which swaps occurred at lag 1 significantly more than at all other lags (ps < .001), where the proportion of swaps did not differ significantly. A significant group by lag interaction was observed (F_GG(3.68, 77.34) = 7.73, p < .001, η² = .27, 95% CI [.09; .39]), in which the 22q11DS group (M_22q11DS = .252, SE = .032) made significantly more swaps at lag 1 than both the MA-matched (M_MA = .079, SE = .032, p = .001) and CA-matched (M_CA = .052, SE = .032, p < .001) control groups, which did not differ from one another. The proportion of swaps were similar across groups at all other lags.

Figure 4.

Proportion of swaps by lag in the Color attentional blink task. Error bars represent +/− 1 standard error.

3.2.4. Binding errors.

Binding errors (i.e., the proportion of trials in which participants inaccurately reported non-target letters at adjacent temporal positions to the actual target) were analyzed separately for each target. For T1 (see Supplemental Figure 4), A RM ANOVA revealed a main effect of group (F(2, 402) = 21.45, p < .001, η² = .1, 95% CI [.05; .15]), in which the CA-matched control group (M_CA = .007, SE = .002) made significantly fewer overall binding errors than both the MA-matched control group (M_MA = .024, SE = .002, p < .001) and the 22q11DS group (M_22q11Ds = .022, SE = .002, p < .001), which did not differ significantly from one another (p = .48). There was also a main effect of position of the reported distractor from the actual T1 (+/− 5, 4, 3, 2, or 1; F_GG(3.62, 1455.74) = 140.1, p < .001, η² = .26, 95% CI [.22; .29]), in which participants reported the distractor directly after (+1) the actual T1 more than distractors in any other position (ps < .001). After position +1, the next highest positions in which distractors were inaccurately reported as T1 were at positions −1 and +2, which did not differ significantly from each other (p = .27), but had higher proportions than further positions in either direction of T1 (ps < .002). Positions −5, −4, +3, +4, and +5 all had similar occurrences of binding errors, while positions −3 and −2 had significantly fewer binding errors than all other positions (ps ≤ .04) except position +4 (ps > .09) and position +5 (−3 only, p > .07). A group by position interaction was observed (F_GG(7.24, 1455.74) = 26.4, p < .001, η² = .12, 95% CI [.08; .14]), in which the MA-matched control group exhibited significantly more binding errors than the CA-matched control group at all positions (ps < .008), as well as more binding errors than the 22q11DS group at positions −5, −4, −3, −2, and +5 (ps < .05), while there were no differences between these two groups at positions −1, +2, +3, or +4. In addition, the 22q11DS group exhibited more binding errors than both control groups at position +1 (ps < .001) as well as more binding errors than the CA-matched control group at positions −1, +2 and +3 (ps < .04), but not at positions −5, −4, −3, −2, +4, and +5.

For T2 (see Figure 5), A RM ANOVA revealed a main effect of group (F(2, 402) = 37.72, p < .001, η² = .16, 95% CI [.10; .22]), in which the CA-matched control group (M_CA = .011, SE = .002) made fewer overall binding errors than both the MA-matched control group (M_MA = .020, SE = .002, p < .001) and the 22q11DS group (M_22q11DS = .031, SE = .002, p < .001), and the MA-matched control group made fewer binding errors than the 22q11DS group (p < .001). There was also a main effect of position of the reported distractor from the actual T2 (F_GG(2.66, 1069.69) = 198.6, p < .001, η² = .33, 95% CI [.08; .27]), in which participants reported the distractor directly after (+1) the actual T2 more than distractors in any other position (ps < .001). After position +1, the next highest positions in which distractors were inaccurately reported as T2 were at position −1, which had higher proportions of binding errors than all other positions (ps < .005), except position +1. Positions −2 and +2 had the next highest proportion of binding errors and did not differ significantly from each other, but were higher than all positions further out (ps < .05), except position +5 (ps > .06) and −4 (−2 only, p > .09). Proportion of binding errors did not differ significantly across positions −5, −4, −3, +3, +4, and +5. A group by position interaction was observed (F_GG(5.32, 1069.69) = 21.12, p < .001, η² = .1, 95% CI [.06; .13]), in which the 22q11DS group exhibited significantly more binding errors than the CA-matched control group at all positions (ps ≤ .02), as well as more binding errors than the MA-matched control group at positions −1 and +1 (ps < .001), but not other positions. In addition, the MA-matched control group exhibited more binding errors than the CA-matched control group at positions −5, −2, +2, +3, +4, and +5 (ps < .05), but not at positions −4, −3, −1, and +1.

Figure 5.

Binding errors by distractor position from the second target (T2) only in the Color attentional blink task. Error bars represent +/− 1 standard error.

3.2.5. Depth of the AB and rate of recovery.

There was a main effect of group (F(2, 42) = 8.37, p = .001) for depth of the AB (see Supplemental Figure 5) and post hoc analysis conducted with Bonferroni corrections revealed that the 22q11DS group (M_22q11DS = .368, SE = .035) exhibited a significantly deeper blink than both the MA-matched control group (M_MA = .168, SE = .049, p = .003) and the CA-matched control group (M_CA = .172, SE = .033, p = .003). There was no significant difference in the depth of the AB between the two control groups (p = 1.0). There was also a main effect of group (F(2, 42) = 31.44, p < .001) for rate of recovery from the AB and post hoc analysis conducted with Bonferroni corrections revealed that, on average, the 22q11DS group (M_22q11DS = 5.47, SE = .29) recovered from the AB at later lags compared to both the MA-matched controls (M_MA = 3.53, SE = .24, p < .001) and CA-matched controls (M_CA = 3.13, SE = .09, p < .001). No differences in recovery rate were found between the two control groups (p = .63).

3.3. Relationships between AB performance and schizophrenia symptoms.

For the participants with 22q11DS, two-tailed Spearman rank correlations were run between SOPS Positive Symptoms Subscale scores (M = 1.53, SD = 2.07) and measures of the AB in the Category and Color Tasks, including depth of the AB and rate of recovery from the AB. No significant correlations were found.

4. Discussion

4.1. Developments in the Understanding of Temporal Attention Processing in 22q11DS

The findings from this study reveal that AB performance in individuals with 22q11DS relative to TD controls is particularly sensitive to differing task demands. On the Color Task, which requires binding of two features (color and letter) in order to accurately report targets, individuals with 22q11DS differed from CA and MA-matched controls on all AB performance measures, exhibiting lower T1 and T2|T1 accuracy rates, a higher proportion of swaps, a higher rate of temporal binding errors, and a deeper and more protracted AB. In contrast, on the Category Task, individuals with 22q11DS performed similarly to CA and MA-matched controls on all measures of the AB, with the exception of overall lower T1 accuracy relative to both comparison groups. These results suggest that attentional processes that involve temporal binding may be significantly affected in individuals with 22q11DS, a finding that importantly, has been previously demonstrated among individuals with schizophrenia (Parsons et al., 2013). However, in individuals with 22q11DS, other aspects of temporal attention processing may be on par with both chronological age and developmental level.

Results of the current study provide new insights regarding attentional functioning in 22q11DS. First, results of the Category Task suggest that some elements of temporal attention processing are relatively intact in individuals with 22q11DS compared to both chronological and mental age-matched TD peers. Although the task was complex, involving several distractor stimuli presented over very short (135ms) intervals, individuals with 22q11DS were able to achieve similar rates of T2|T1 accuracy as both TD control groups. However, individuals with 22q11DS did exhibit lower rates of T1 accuracy compared to both CA and MA matched control participants in the Category Task. Given similar levels of conditional T2 accuracy rates between groups, it could be the case that the approximate 10% difference in T1 accuracy reflects difficulty in attentional selection in a manner similar to patients with schizophrenia. That is, participants with 22q11DS were less able to correctly bias their attentional selection to the first target but on trials where they did correctly select the first target, they were as able as peers at locking on to, and reporting both of the correct targets.

Results of the Color Task provide insight into an area of attentional weakness that has not yet been explored in 22q11DS. In this task, individuals with 22q11DS performed significantly worse on all measures, showing lower rates of T1 and T2|T1 Accuracy, as well as a deeper and more protracted AB compared to both TD control groups. These results appear to be at least partially contingent on the temporal binding demands of this particular task, which required participants to “bind” the defining feature of targets (in this case, the color purple) to the appropriate letters that were presented rapidly in short (135ms) temporal windows.

Although there was a differential pattern of binding errors as a function of target, with a higher proportion of binding errors made at more distal positions from the target by the younger control group for the first target (T1) and by the 22q11DS group for the second target (T2) in the RSVP stream, the key finding from the binding data relates to the very high proportion of errors made at position +1 for the participants with 22q11DS. While TD participants were also most likely to make binding errors at position +1 from the target, those with 22q11DS did so on almost 15% of trials, more than double that of the TD control participants. This might suggest a slightly slowed processing speed in individuals with 22q11DS, or could potentially reflect a longer temporal window of integration, as has been reported in individuals with schizophrenia (Parsons, et al, 2013). However, this needs to be examined more thoroughly before firm conclusions can be drawn. As no studies of the AB in schizophrenia have used color-based tasks, direct comparisons cannot be made.

Although we believe that impaired temporal binding processes in 22q11DS best explain our findings, alternative explanations cannot be ruled out. For example, it could be that the colors purple and black (as opposed to colors with a greater degree of visual contrast such as yellow and black) were difficult for all participants to discriminate, resulting in reduced accuracy on the Color Task compared to the Category Task. However, if this were the case, control participants would likely have shown a greater decrease in accuracy across lags, rather than just at lag 1. Notably, accuracy at lag 1 was decreased in all groups in the Color Task relative to the Category task, and the control groups failed to demonstrate a phenomenon known as lag 1 sparing. Lag 1 sparing is characterized by spared accuracy at lag 1 relative to the attentional blink period which begins at lag 2 (Potter, Chun, Banks, & Muckenhoupt, 1998) and is often observed in categorical AB tasks. However, lag 1 sparing has not been observed in feature-based AB tasks involving color (Russo, Kates & Wyble, 2017). This is likely due to the increased difficulty in detecting and binding the defining feature (in this case, the color purple) to the correct letters of two targets when they are presented directly next to each other (as opposed to when there are black distractors in between them as there are at subsequent lags). This would explain the lack of lag 1 sparing in the control groups observed during the Color Task, while their accuracy increased to similar levels as observed in the Category Task at later lags.

In contrast, the 22q11DS group showed spared accuracy at lag 1 in the color task, which is not typically observed. This may have been facilitated by the differential pattern of binding errors observed in this group, as they were significantly more likely to bind the color purple to letters directly preceding or following the target. This binding difference may have been beneficial in obtaining higher T2|T1 accuracy at lag 1 in the Color Task since the two purple targets were presented in direct succession. In fact, temporal order swaps (which were still counted as accurate) were much higher in the 22q11DS group specifically at lag 1. At other lags, the tendency to bind the color purple to letters directly before or after the purple target would result in much lower T2|T1 accuracy, as observed, since in these cases the surrounding letters were black distractors rather than other purple targets. One possible explanation of these group differences in T2|T1 accuracy is that temporal binding windows (i.e. the period of time required to accurately bind the features of targets) may differ in these groups. For example, the control participants may experience a shorter temporal binding window such that they reach high levels of accuracy by lag 3, while the 22q11DS group may experience a longer temporal binding window and are unable to achieve high levels of accuracy until lag 5, when the two purple target are much more temporally distal. Although we believe that our findings provide considerable evidence that the differential performance of individuals with 22q11DS on the Color Task compared to the Category Task is due to impairment in typical temporal binding processes, further research is needed to replicate these findings and to demonstrate whether these results are consistent when employing a variety of stimuli, including color targets that vary in their degree of visual contrast to distractors.

While there were no significant relationships observed between positive prodromal symptoms of schizophrenia and measures of the AB, it is important to note that none of our participants had a diagnosis of, or met diagnostic criteria for schizophrenia. Few individuals in our sample exhibited prodromal symptoms as measured by the SOPS and only 1 of 15 obtained a score of 3 or above on any item of the Positive Symptoms Subscale, which is considered the threshold for clinically significant, prodromal symptoms of psychosis. In addition, the mean age of our 22q11DS sample was younger than 21 years old, while schizophrenia is most typically diagnosed in the early to mid-twenties (Beiser, Erikson, Fleming, & Iacono, 1993; Manschreck, Maher, & Candela, 2004), suggesting, perhaps, that symptoms may not yet have fully manifested. Finally, the participants in this study were part of a much larger, longitudinal study of the development of 22q11DS. The two tasks presented here were completed during the fourth wave of data collection, and several participants who had already converted to schizophrenia were too psychiatrically impaired to travel to our center to participate in this timepoint. Accordingly, the lack of relationship between the AB and measures of prodromal symptoms might be due to the restricted range of prodromal scores in the sample, rather than a lack of relationship between the AB and schizophrenia symptoms in this population.

4.2. Clinical Implications

Results of the current study yield important clinical considerations. This study is the first to our knowledge to provide evidence of temporal binding impairments in individuals with 22q11DS, which may have real world implications for these individuals. For example, researchers (Parsons et al., 2013) have suggested that the perceptual processing differences related to temporal integration found in individuals with schizophrenia may account for higher level impairments, such as a disrupted sense of continuity and fragmentation of events in time that many patients with schizophrenia report. It is possible that low-level perceptual processing differences may impact higher-order cognitive and executive processes in individuals with 22q11DS in various ways. In addition, reductions in T1 accuracy rates of those with 22q11DS relative to control groups on both AB tasks may be indicative of difficulties in sustaining attention in this population. This finding highlights a potential mechanistic link to ADHD, another psychological disorder that is highly prevalent in individuals with 22q11DS compared to the general population (Schneider et al., 2014). It will be important to continue investigating temporal attention processing in individuals with 22q11DS and to consider the consequences in the context of psychological assessment and treatment.

4.3. Limitations and Future Directions

Limitations of the current study include a small sample size of individuals with 22q11DS with limited ranges in chronological age and developmental attainment. In addition, although we have attempted to relate the functional abilities of individuals with 22q11DS in the domain of temporal attention processing and temporal binding errors to what we know about these processes in schizophrenia, no direct comparisons can be drawn since we have not examined non-syndromal individuals with schizophrenia and our methods do not exactly match studies that have. Future research should explore the AB in younger individuals with 22q11DS, examine temporal binding in 22q11DS in different experimental paradigms, examine performance on Color AB tasks in schizophrenia patients, and directly compare performance of individuals with schizophrenia and 22q11DS on AB tasks and other tasks that measure temporal binding and temporal attention processing to better understand the overlaps and nuances that may exist between these comorbid groups.