Results
A total of 55 adolescents were included in these analyses; however, not all participants completed both antisaccade and PPI components of the study. As such, sample size varied depending on the analysis. The groups (22q11DS/HC) did not differ from each other in age (t(53) = 0.28, p = 0.78) or gender (χ2(1) = 0.25, p = 0.62), but IQ difference was greater than two standard deviations between groups (Full-scale IQ: mean (SD): 22q11DS = 75.4(15.3); HC = 104.4(15.8), Table  1).
Table 1  Neuropsychological characteristics of the 22q11DS and HC groups *p < 0.05; **p < 0.001.

Pre-pulse inhibition (PPI) task performance
Startle eyeblink responses were modulated by pre-pulse interval (120/480 ms: F(1,34) = 9.84, p = 0.004) and by condition (passive/active: F(1,34) = 35.98, p < 0.001). Both groups showed PPI to the 120-ms pre-pulse interval versus baseline startle response (no pre-pulse presented) in both the passive and active pre-pulse conditions (passive task: t(18) = 2.72, p = 0.014 HC subjects; t(16) = 5.35, p < 0.001 22q11DS subjects; active task: t(18) = 4.30, p < 0.001 HC subjects; t(16) = 2.58, p = 0.02 22q11DS subjects, Table  2).
Table 2  Electromyographically recorded startle eyeblink responses at baseline (no pre-pulse presented) and following presentations of subtle acoustic pre-pulses at 120- and 480-ms lead intervals Resulting percent change as pre-pulse inhibition (%PPI) or pre-pulse facilitation (%PPF) relative to baseline startle response (standard error mean in parenthesis).
The three-way interaction of pre-pulse interval by attention conditions by group approached significance (F(1,34) = 3.83, p = 0.059; Figure  1). Post hoc testing revealed a differential modulation of sensorimotor gating between groups. HC participants showed a significant increase in PPI in the 120-ms pre-pulse interval when directing attention to the pre-pulse, compared to the passive listening task (t(18) = -2.42, p = 0.026). In contrast, 22q11DS subjects showed a trend towards reduced PPI when directing attention to the pre-pulse (t(16) = -1.80, p = 0.09) at the 120-ms pre-pulse interval compared to the passive listening task. Moreover, 22q11DS subjects showed significantly greater PPF in the passive listening task at the 480-ms pre-pulse interval compared to the active listening task (t(16) = -2.31, p = 0.035). Figure  1 indicates both groups exhibit PPF; however, this differs between conditions, with the HC group exhibiting greater (although statistically non-significant) PPF during the active condition.
Figure 1  Mean percent pre-pulse inhibition or pre-pulse facilitation for 120- and 480-ms pre-pulse lead intervals. Mean percent pre-pulse inhibition (PPI) or pre-pulse facilitation (PPF) for 120- and 480-ms pre-pulse lead intervals relative to baseline startle response (no pre-pulse presented). Note differential attention effects on PPI and PPF between groups depending on performance on an auditory discrimination task on the pre-pulse (active) and not (passive).
The ANOVA revealed no additional significant interactions with none of the following conditions reaching significance: interval by group (F(1,34) = 0.045, p > 0.05); condition by group (F(1,34) = 0.08, p > 0.05) nor interval by condition (F(1,34) = 0.00, p > 0.05).

Antisaccade task performance
Group comparisons on antisaccade error, latency and amplitude showed the 22q11DS group made significantly more antisaccade errors compared to the HC group (F(1,43) = 28.93, p < 0.0001) but did not differ on measures of latency or amplitude (p > 0.05). These effects remained non-significant when separated by trial accuracy (correct/incorrect antisaccade). However, antisaccade error was influenced by response latency with increased latency for successful antisaccade inhibition trials in both groups (F(1,41) = 41.63, p < 0.0001) (Table  3).
Table 3  Mean (SD) antisaccade parameters for 22q11DS and HC groups ***p < 0.0001.

Associations amongst antisaccade, PPI and executive functioning
Statistical significance of the difference between correlation coefficients for the 22q11DS and HC groups was examined because the patterns of association differed between the groups on several measures. Where the groups did not differ significantly (z(obs): -1.96–1.96), we performed partial correlations controlling for group membership to explore the relationship between the key variables of interest. Bonferroni-adjusted significance level of 0.004 was calculated to account for the increased possibility of type-I error. Following Bonferroni correction, antisaccade error was moderately associated with executive function measures of planning ability, with increased moves on the Tower of London task associated with increased rates of antisaccade error (df = 37, r = 0.447, p = 0.004). Zero-order correlations indicated a moderate effect of group (r = 0.630).