|Home | About | Journals | Submit | Contact Us | Français|
The goals of the study were to examine inhibitory deficits on the antisaccade task in 8- to 20-year olds with youth-onset psychosis or attention deficit/hyperactivity disorder (ADHD) and healthy controls and to examine if age-related changes in performance differed across groups. In addition to the conventional measures of performance, pupillary dilations were used to obtain estimates of phasic and tonic level of arousal. Results showed that the psychosis, but not the ADHD, group had elevated antisaccade error rates; however, variability of error rates was high in all groups. These inhibitory failures were accompanied by a lower level of momentary cognitive effort (as indexed by pupillary dilations). The largest differences between the control and clinical groups were found not in the expected indices of inhibition but in the probability of correcting inhibitory errors and in variability of antisaccade response times, which were correlated with each other. These findings did not appear to be attributable to a deficit in maintaining task instructions in mind in either disorder or lack of motivation in ADHD. Instead, results point to impairments in both clinical groups in sustaining attention on a trial-by-trial basis, resulting in deficits in self-monitoring. Thus, results show inhibitory deficits in the context of more general attentional impairments in both disorders.
There are intriguing similarities between schizophrenia and attention deficit/hyperactivity disorder (ADHD), including impairments in the same general cognitive domains, behavioral symptoms such as inattentiveness, and abnormalities in the same general neurotransmitter systems and brain regions.1,2 Yet, there have been relatively few direct comparisons between schizophrenia and ADHD. Most of these comparisons have involved youth-onset schizophrenia, a rare, severe, and more genetically loaded form of the disorder that does not differ qualitatively from the adult-onset form.3–11
One of the cognitive domains that is impaired in both schizophrenia and ADHD is inhibition. Thus, the goal of the current study was to use saccadic and pupillary dilation measures to examine inhibitory impairments on the antisaccade task in 8- to 20-year olds with youth-onset psychosis or ADHD and healthy controls.
On the antisaccade task, participants are presented with a visual stimulus on one side of a computer screen. They are instructed to look away from it to a blank location equidistant from the center but on the opposite side. Performance on this task is compared with that on the prosaccade task, where participants are instructed to look at the target. In adults, typical saccadic response times (RTs) are 150–250 milliseconds on the prosaccade task and longer by 100–200 milliseconds on the antisaccade task.12 Error rates are variable, and even healthy adults make many errors depending on the task parameters. The vast majority of these errors are corrected,13 although not necessarily recognized.14,15
The cognitive processes involved in the antisaccade task likely include the following (not necessarily in sequential order): keeping task instructions in mind, maintaining preparedness to respond, detecting the target, inhibiting the automatic saccade and refraining from prematurely disengaging visual-spatial attention from fixation, moving attention covertly to the opposite side, computing the transformed sensorimotor coordinates for the internally generated antisaccade, and executing the antisaccade.16–19 Error rates on this task reflect the ability to inhibit, and saccadic RTs reflect the total duration of the cognitive and motor processes on trials on which the automatic saccades are inhibited successfully.
Only 3 studies have investigated visually guided saccades in youth with, or at risk for, schizophrenia. One study20 found hypometric saccades in children at genetic risk for schizophrenia. In contrast, an early study found no significant differences between teenage children of mothers with schizophrenia and healthy controls in either RT or amplitude of visually guided saccades.21 A more recent comparison of participants with childhood-onset schizophrenia and healthy controls also showed no significant differences in RT of visually guided saccades.22
Pro- and antisaccades have been examined extensively in adults with schizophrenia-spectrum disorders.23–27 Most studies show no impairments in amplitude, peak velocity, gain, RT, or duration of visually guided saccades. In contrast, results consistently show increased antisaccade error rates in schizophrenia, with less consistent results for RT, amplitude, peak velocity, and duration of antisaccades. Indeed, Turetsky et al27 report that in more than 50 published studies of the antisaccade task in schizophrenia, all studies have found elevated error rates in schizophrenia. Studies generally report that the majority of erroneous antisaccades are corrected as evidence that the schizophrenia group was motivated to perform the task; however, rates of corrective saccades are usually not compared directly between schizophrenia and control groups.
Most studies of pro- and antisaccades in children and adolescents with ADHD have reported elevated antisaccade error rates in ADHD.28,29 Prosaccade RTs have not been found to be elevated in most studies. Results for antisaccade RTs are mixed. Variability of pro- and antisaccade RTs was measured in 2 studies, and both found increased variability in ADHD compared with control samples. Corrective saccades were also found to be reduced in ADHD in both the studies in which they were measured (reviewed in Karatekin29).
There have been 3 studies in which adults with ADHD were compared with healthy controls on pro- and antisaccade tasks.30–32 All 3 studies found elevated rates of antisaccade errors in ADHD. Results have been mixed for pro- and antisaccade RTs. Variability of RTs and frequency of corrective saccades were not measured in any of the studies with adults.
Hypotheses regarding antisaccade deficits in ADHD include prefrontal and/or frontostriatal dysfunction,31,33–36 difficulty keeping task instructions in mind,34 a weak fixation system,36 impairments in motor inhibition,32,34 and generation of internally guided responses.34 Antisaccade impairments could also be related to impairments in moving covert visual-spatial attention.21,37,38
One source of inhibitory failures could be insufficient recruitment of cognitive resources, which can be operationalized with a psychophysiological measure assessing arousal: pupillary dilations. Thus, we measured pupillary dilations to examine the role of tonic and phasic arousal in inhibitory performance. Pupillary diameter shows both tonic and phasic changes.39–42 Task-specific pupillary dilations are phasic changes in pupillary diameter time-locked to the onset of stimuli or responses during cognitive tasks. Phasic pupillary dilations typically occur within 500–1500 milliseconds of stimulus onset, depending on the task. They are mediated through rapid interactions among the frontal cortex, limbic regions, diencephalon, and reticular activating system. The final stages of this process involve 2 pathways terminating in the muscles regulating pupillary diameter. Pupillary dilation in response to cognitive stimuli can result from phasic activation of the sympathetic pathway, which controls pupillary dilation, or phasic inhibition of the parasympathetic pathway, which controls pupillary constriction.40,43,44
Pupillary dilations have not been reported in any studies of youth-onset psychosis or ADHD. However, there is a long line of research demonstrating reduced dilations to cognitively demanding stimuli on a variety of tasks in adults with schizophrenia.42,45–50 Some studies also show abnormally large dilations in schizophrenia, which has been interpreted as indicating that individuals with schizophrenia spend more effort on automatic processes that produce minimal dilations in controls.51
In a companion study examining factors affecting antisaccade performance in ADHD,28 we found that adolescents with ADHD were impaired relative to age-matched controls in accuracy and saccadic RT on the first administration of a 32-trial version of the antisaccade task. Although their accuracy improved on the second administration of the same task at the end of the session, it did not improve disproportionately compared with controls. In contrast, the saccadic RTs of the ADHD group came close to normal during the second administration, indicating that elevated antisaccade RTs in this disorder may be related to regulating arousal on a novel task. The ADHD group also made fewer corrective saccades than both the age-matched and younger groups, suggesting difficulties with impulsivity and goal neglect. Because of the small number of trials and the small sample size, we did not examine variability in performance or pupillary dilations.
In the current study, we address questions stemming from this companion study and extend the findings to psychosis. To our knowledge, neither these 2 disorders have been compared directly on these tasks nor have pupillary dilations ever been examined during pro- and antisaccades. The first aim was to test if the clinical groups would show elevated antisaccade error rates and to compare the extent to which they would correct their errors. A reduced frequency of corrective saccades would indicate difficulties in self-monitoring. Based on the bulk of the evidence, we expected increased antisaccade errors in both disorders and fewer corrective saccades at least in ADHD. Second, we examined pupillary dilations to test if inhibitory failures were associated with difficulties in phasic resource recruitment. We also examined absolute pupillary diameters to examine the role of tonic arousal level on inhibitory performance and to determine if group differences in pupillary dilation were due to differences in baseline pupillary diameter. Based on previous research, we expected reduced dilations in the psychosis group on the antisaccade task but made no predictions for the ADHD group. Third, we measured latency to initiate saccades and to reach peak pupillary dilation to test if participants were slow to process information on correct and incorrect trials and variability of saccadic RTs to test if inhibitory failures were associated with attentional fluctuations. Previous research led us to predict normal prosaccade RTs in both disorders but more variable RTs at least in ADHD. Finally, we examined pupillary waveforms to explore differences in the temporal course of information processing on pro- and antisaccade trials.
Table 1 lists participants’ demographic and clinical characteristics. Participants with psychosis were recruited from inpatient and outpatient clinics at the University of Minnesota, mental health professionals in the community, and flyers distributed at regional mental health conferences. One participant was in residential treatment at the time of the study. The rest were living at home. Participants in the control and ADHD groups were recruited from advertisements in the local community and friends of other participants. ADHD participants were also recruited from support groups for ADHD.
Potential participants were excluded if they were not fluent in English or were color blind, if they had been premature by more than 4 weeks, if they had a history of significant neurological conditions, or if they had an IQ lower than 70. Potential participants were excluded from the ADHD and control groups if they had been adopted or had a first-degree biological relative with schizophrenia. Potential participants were excluded from the ADHD group if they were taking psychoactive medications other than psychostimulants, if their parents were not willing to discontinue psychostimulants for 24 hours prior to cognitive testing, if they had been diagnosed with or suspected of having a pervasive developmental disorder, or if they had never met criteria for the combined subtype. We included 2 adolescents who currently met criteria for the inattentive subtype but who had previously met criteria for the combined subtype and scored above 60 on the Attention Problems Scale of the Achenbach Child Behavior Checklist.58 Potential controls were excluded if they had ever taken psychoactive medications, been diagnosed with a major psychiatric disorder or met criteria for a current disorder, had attention problems for which they had sought help, or had first-degree biological relatives with ADHD.
Diagnoses were made using Diagnostic and Statistical Manual of Mental Disorders (Fourth Edition) criteria59 and based on semistructured interviews (Kiddie Schedule for Affective Disorders and Schizophrenia-Present and Lifetime Version)60 conducted separately with participants and at least one parent/guardian.
Diagnoses are listed in table 2. In the psychosis group, age of onset of psychotic symptoms averaged 12.9 years (SD=3.0, range=7–17). Participants with psychosis not otherwise specified (NOS) were included only if they had biological relatives with schizophrenia. Four of the participants diagnosed with psychosis NOS had first-degree biological relatives with schizophrenia, and one had a second-degree biological relative with schizophrenia and a biological sibling with psychotic symptoms.
Families were provided with monetary compensation for participation. Most families were also provided with a diagnostic report.
Participants were asked to refrain from taking psychostimulants for at least 24 hours prior to testing. However, on the day of testing, one participant in the psychosis group had taken a psychostimulant. In the psychosis group, 2 of the participants on antipsychotic medications were taking typical neuroleptics (molindone and haloperidol). The other participants were all on atypical neuroleptics.
Other details of the diagnostic procedure can be found in (Karatekin et al61).
The same stimuli were presented in the same order to all participants in both the pro- and antisaccade conditions. Each trial began with a black fixation cross at the center of the screen. A small visual stimulus (12.7 × 9.5 mm or 1.1° × 0.8° of visual angle) appeared for 1500 milliseconds 11.8 cm (9.7°) to the right or left of center simultaneously with the disappearance of the cross. The stimulus was the picture of an ant in the prosaccade condition and a bug in the antisaccade condition. Stimulus location was determined randomly on each trial, with the constraint that it appeared equally often on each side. In the prosaccade condition, participants were instructed to look at the ant as soon as it appeared. In the antisaccade condition, they were instructed to look in the opposite direction, “where the bug would have appeared if it were on the other side.” Verbal instructions were complemented with visual instructions regarding the saccade goal. Instructions in both conditions emphasized both speed and accuracy. No instructions were given regarding correction of errors.
All stimuli were presented against a white background on a Dell Trinitron VGA color monitor (53 cm diagonal, resolution: 640 × 480 pixels). The stimuli remained on screen for 1500 milliseconds, and the intertrial interval alternated randomly among 1000, 1500, and 2000 milliseconds on both conditions. During the intertrial interval of both conditions, participants were shown a white screen with the fixation cross in the center and instructed to look at the cross. Participants were administered 10 practice and 96 experimental trials in both conditions. Practice trials were repeated if necessary. Feedback was not provided.
The current study was part of a larger study of control over attention and goal-directed action in youth-onset psychosis and ADHD. As part of this larger study, participants were administered other tasks over 2 sessions, and some underwent neuroimaging (6 control and 9 psychosis participants were included in White et al62; 45 controls were included in Karatekin et al63; 46 control, 26 ADHD, and 29 psychosis participants were included in Karatekin et al61). The current task was always administered on the first day of testing. Most of the participants were administered a divided attention task and a serial RT task during the same session as the pro- and antisaccade conditions. The divided attention task was always presented first. The pro- and antisaccade and serial RT tasks were administered next in counterbalanced order, and the pro- and antisaccade conditions were presented in counterbalanced order. Each condition took approximately 5–6 minutes to administer. Participants were provided with rest breaks between tasks and conditions as necessary.
Horizontal and vertical coordinates of the center of gaze and pupillary diameter were recorded with a video-based eye monitor (ISCAN Eye Tracking Laboratory, Model ETL-400, Woburn, MA), which has a temporal resolution of 60 Hz and a spatial resolution of 1° over the range of visual angles used in the present study. The spatial resolution for measurement of pupillary diameter was 0.037 mm.
Participants sat 69 cm from the monitor on which the stimuli were displayed. A camera with an attached infrared light source to illuminate the pupil was positioned in front of the computer screen on which the stimuli were presented below eye level and 40 cm from the participants’ eye. The camera recorded the movements of the participants’ left eye. Because the camera automatically compensated for small head movements, participants’ heads were not restrained. However, they rested their heads against a padded head rest. The experimenter sat behind the participant in front of the computer that controlled the video camera and collected the eye movement data.
Gaze position was calibrated for each participant at the beginning of the session by focusing the camera on his/her left eye and having him/her look at small visual stimuli in the center and 4 corners of the screen. These positions were recorded as the targets of gaze. Calibration was repeated between blocks of trials if necessary due to excessive head movement.
Custom software was used to present the stimuli, and the eye-tracking data were merged offline with time-linked stimulus presentation data.
The experiment was conducted in a windowless room with fluorescent lighting. The luminance of the screen on which the stimuli were displayed was 140 cd/m2, and the average luminance of the white wall that was directly behind the computer was 139 cd/m2, as measured with a spot photometer. Assuming a high reflectance for the wall (0.7–0.8), we can estimate the ambient illumination in the room to be approximately 550–625 lux.64(box 4.3) Because the target stimulus was only 1.1° × 0.8°, there was not a significant change in luminance when it was presented on the screen.
Custom software was used to extract saccades from the raw data. A saccade was defined as an eye movement with a velocity between 90°/s and 800°/s over at least 33.3 milliseconds. Saccadic RT was specified as the time between stimulus onset and saccade onset. Saccades that occurred less than 50 milliseconds after an artifact (as defined under “Analysis of Pupillary Dilations”) were excluded. Saccades that occurred less than 83 milliseconds after stimulus onset were excluded as anticipatory responses. Saccades that occurred more than 800 milliseconds after target onset were excluded as being too slow. Saccadic RTs longer than 3 SDs above the participant's own condition mean for valid saccades (ie, nonpremature saccades to the correct location) were also excluded as outliers. Across both conditions and all participants, the maximum number of trials eliminated due to outliers ranged from 4 to 5. Median saccadic RT data, measured separately for each visual stimulus, were calculated for correct and incorrect trials in each condition. Variability of saccadic RTs was estimated with the coefficient of variation (SD/mean).
The landing position of the saccade was defined as the horizontal and vertical coordinates of the gaze 67 milliseconds after the onset of the saccade. To correct for minor head movements, the location of the gaze was corrected for initial fixation position (average location of gaze during the 83 ms before target onset). Trials with initial gaze positions more than 4.2° from the calibrated center were excluded.
A saccade was counted as correct if it landed more than 1.05° away from fixation and on the side of the visual stimulus in the prosaccade condition and on the side opposite the stimulus in the antisaccade condition. Erroneous antisaccades (or antisaccade errors) were defined as saccades in the antisaccade condition that were directed at the visual stimulus instead of the opposite direction. If a participant's first valid saccade was incorrect, the landing position of the second saccade was analyzed to determine if it was a corrective saccade. A saccade was counted as a corrective saccade only if its onset occurred 83 milliseconds after the onset of the previous saccade in order to avoid defining movement-related artifacts occurring right after a first saccade as the onset of a new saccade. RTs of corrective saccades were defined as the time between the onset of the corrective saccade and the onset of the initial erroneous saccade.
The proportions of correct and incorrect saccades were calculated by dividing the number of trials with correct or incorrect saccades, respectively, by the total number of trials on which a saccade was detected in that condition. The proportion of corrective saccades was calculated by dividing the number of trials with corrective saccades by the number of incorrect trials in each condition.
To calculate pupillary dilations, we first used an algorithm to identify blinks and saccade-related artifacts in the pupillary record. Blinks and artifacts were defined as (a) the pupil diameter falling below 1.86 mm or above 5.96 mm, (b) the horizontal or vertical positions of the eye falling outside the limits of the screen, (c) the diameter of the pupil changing by more than 0.75 mm over 16.7 milliseconds, or (d) velocity of the eye movement between 2 consecutive records exceeding 80°/s. When blinks/artifacts were detected, an algorithm used pre- and postblink/artifact values to perform a linear interpolation of pupillary values throughout the duration of the blink/artifact, starting 34 milliseconds before and ending 34 miliseconds after the identified blink/artifact. Pupillary data were discarded for blinks/artifacts lasting longer than 500 milliseconds.
Pupillary dilation was defined as the magnitude of the adjusted maximum pupillary dilation during the 800 milliseconds after the onset of the stimulus (the 800-ms value was arrived at after inspection of the pupillary waveforms). To measure pupillary dilation, we subtracted pupillary diameter at each 17-milisecond interval from the baseline diameter for that stimulus. Baseline was defined as average pupillary diameter during the 100 milliseconds preceding each stimulus. These baseline values were also used as the measure of absolute pupillary diameter, averaged across correct and incorrect trials. Instead of using the observed peak value of the resulting pupillary waveform to calculate the magnitude of pupillary dilation, we used a least squares method to fit a quadratic curve to the data in the neighborhood of the peak value. In fitting this curve, the values at the peak and 67 miliseconds before and after the peak were used. The maximum point of this quadratic curve was used as an adjusted peak. The magnitude of the pupillary dilation was defined as the height of the adjusted peak. Because this value was based on 9 data points, this procedure was expected to yield a more stable estimate than the height of the observed peak as measured by 1 data point. We did not take the average diameter value over a prespecified time after stimulus onset because this method could have confounded the time to reach peak pupillary dilation with the peak itself.
To remove blink- and saccade-related artifacts from the data that may have been missed by the algorithms, we excluded from analyses all adjusted pupillary dilation values that were smaller than −0.11 mm or larger than 1.12 mm. These absolute threshold values for outliers were based on visual inspection of all the participants’ data, without regard to condition. Adjusted pupillary dilations were calculated separately for each saccade type (correct pro- and antisaccades, erroneous antisaccades). Median pupillary dilations were used in the analyses to minimize the effect of outliers.
Latency to peak dilation was defined as the difference in time between stimulus onset and peak pupillary dilation. Peak latencies that occurred less than 200 milliseconds after stimulus onset were not analyzed. Median latency data, measured separately for saccade type, were calculated in each condition.
Pupillary waveforms were created by averaging pupillary dilations during the 800 milliseconds after stimulus onset for correct and incorrect trials within each participant and then averaging the data across participants within each diagnostic group and condition.
There were not enough trials with uncorrected antisaccade errors to yield reliable data, so to reduce variability, pupillary dilations, latency to reach dilation, and pupillary waveforms for erroneous saccades were calculated only for those trials on which a corrective saccade was made.
The criteria for defining blinks/artifacts, saccadic RTs, landing position of saccades, and pupillary dilations were arrived at in previous studies in the same laboratory by visually examining eye movement records of healthy young adults and 10-year olds to determine the optimum criteria for distinguishing between legitimate saccades and artifacts. Almost all these criteria were identical to those used on other tasks in our previous studies of healthy individuals and 10-year olds.65–68
Statistical analyses were conducted with SPSS 14.0 and MacAnova 5.06 (an open-source cross-platform statistics program available for Windows, Macintosh, and Linux at http://www.stat.umn.edu/macanova/).
Appropriate transformations of responses to achieve normality and constant variance were sought among the Box-Cox family of distributions. These are equivalent to power transformations y → yP, except that y → log(y) when p=1. The power was selected to be close to the maximum likelihood estimate of p using a graphical procedure.69 Through this procedure and by visual inspection of the data, an arcsine transformation was used on the proportion of trials with detected, correct, and corrective saccades; logarithmic transformations were used on median pupillary dilations, latencies and variability of saccadic RTs, latencies to reach peak pupillary dilation, and anticholinergic equivalents of antipsychotic medications; and a square root transformation was used for chlorpromazine equivalents of antipsychotic medications.
Continuous demographic variables were analyzed with univariate analyses of variance (ANOVAs), and significant findings were followed up with Tukey tests. Categorical demographic variables were analyzed with chi-square tests. Correlations between variables within each group were calculated using Pearson correlation coefficients, controlling for age and IQ except when noted otherwise.
Repeated-measures type-III analyses of covariance (ANCOVAs), with age as the covariate, were used to examine effects of subject variables. Type-III sums of squares were used because they test the same hypotheses as post hoc tests computed from the same ANOVA model. The covariate was modified by subtracting the mean age of all participants from each participant's age. Each ANCOVA tested linear and quadratic trends for age. We examined quadratic trends in addition to linear trends because age trends were clearly nonlinear in some cases, and a quadratic trend is among the simpler alternatives to a linear trend. Models were selected by backward elimination of nonsignificant terms involving age, starting with the highest order interactions. When the quadratic trend on age was significant, the linear trend was not reported. IQ was also used as a covariate in the analyses in the same manner as age; however, we assessed only linear trends for IQ. Huynh-Feldt adjustments to dfs were used to compute F statistic P values, and Huynh-Feldt–adjusted dfs were reported where applicable.
Post hoc analyses of ANCOVA results were conducted using custom macros for MacAnova. Main effects or interactions were generally not followed up when there were higher order interactions involving the same variables. Tests of between- and within-subject contrasts and slopes were based on appropriate t statistics. To protect against multiple testing, P values were Bonferroni corrected, ie, multiplied by the appropriate number of simultaneous tests. When the contrast involved a between-subjects contrast, Tukey-Kramer P values based on the Studentized range were computed and then, where appropriate, Bonferroni corrected by the number of intrasubject contrasts being considered simultaneously.
To calculate effect size, we used a measure similar to Cohen d but that took into account the age differences among groups. Specifically, we divided the difference of the group means (age adjusted as appropriate) by the square root of the mean square error term for the between-subjects analysis section of the ANCOVA. In cases where there was an interaction between group and age, the value reflects the size of the effect at the average age for the whole sample. When the 2 groups are similar in age, as in the ADHD-control comparisons, our measure of effect size and Cohen d yield similar results.
All tests were 2 tailed, and findings are reported as significant if α ≤ .05. All significant results of multiway analyses are listed; therefore, if a main effect or interaction is not mentioned in the results, it can be inferred that it was not significant.
We first analyzed the proportion of trials with valid, detected saccades (as defined in the “Methods” section) in the 2 conditions. Saccades might not have been detected on a trial because the participant did not make a saccade or, more likely, blinked or moved his/her head at the time of stimulus onset. The proportion of trials with detected saccades (correct or incorrect) exceeded 75% across all groups and conditions. An ANCOVA, with age and IQ as covariates, showed a linear increase with age, F1,87=5.19, P=.025, and a diagnosis effect, F2,87=13.21, P < .001. The control group had more trials with detected saccades than the clinical groups, who did not differ. There was also an interaction between diagnosis and IQ, F2,87=4.77, P=.011. Higher IQs were associated with more detected saccades in the psychosis group but not in the other 2 groups. The diagnosis by condition interaction was not significant, P=.110, indicating that task difficulty did not differentially affect the proportion of trials with detected saccades across groups.
As can be seen in table 3, there were few prosaccade errors in all groups. Because of floor effects, these data were not analyzed further. On antisaccade trials, participants in all groups made errors on approximately 50% of the trials. However, there was considerable variability in all groups, with SDs of 20%–25%. Error rates decreased linearly with age, F1,99=36.23, P < .001. There was a diagnosis effect, F2,99=3.93, P=.023: The psychosis group made more errors than the controls.
It should be noted that the lack of an age-by-diagnosis interaction in this analysis was not due to low power: Partial η2 was 0.01 for the interaction between diagnosis and the linear trend for age and less than 0.001 for the interaction between diagnosis and the quadratic trend for age.
Most of the antisaccade errors were corrected in all groups, although participants were not given any instructions regarding error correction (see table 3). Corrective saccades increased linearly with age, F1,99=5.98, P=.016, replicating previous studies (Everling and Fischer).17 There was a diagnosis effect, F2,99=16.58, P < .001. Controls made more corrective saccades than the clinical groups, who did not differ from each other.
A comparison of pupillary waveforms for pro- and antisaccades across diagnostic groups are displayed in figure 1. Figure 2 displays the same data within each group to enable direct comparisons across saccade types. The subsequent analyses were focused on the initial peak pupillary dilation during the first 800 milliseconds after stimulus onset.
Pupillary dilations showed an effect of saccade type, F2,200=30.94, P < .001: Dilations were larger for both correct and erroneous antisaccades than for prosaccades. There was also a diagnosis effect, F2,100=4.63, P=.012: Dilations were larger in the control than in the psychosis group. The difference between the ADHD and psychosis groups did not reach significance, P=.062. (Although we attempted to match participants on pupillary dilation in the antisaccade condition to test if group differences in antisaccade errors would remain after matching, this was not possible due to the fact that the majority of the participants in the psychosis group had small dilations and matching would result in too much reduction of statistical power.)
In the control group, higher error rates were strongly correlated with smaller pupillary dilations in the antisaccade condition, r45=−.51, P=.003. In contrast, this correlation was nonsignificant in both the ADHD, r19=.03, P=.911, and psychosis groups, r22=−.22, P=.300 (IQ and age were covaried in these analyses).
Absolute pupillary diameters were larger in the anti- than in the prosaccade condition, F1,99=8.18, P=.005, and diameters decreased linearly with age, F1,99=12.57, P=.001. The diagnosis-by-condition interaction just missed significance, P=.051. In addition, larger pupillary dilations were correlated with larger pupillary diameters in all groups (table 4). Thus, group differences in pupillary dilations cannot be attributed to differences in absolute pupillary diameter.
Saccadic RTs were longer by 120–150 milliseconds for anti- than for prosaccades (see table 3). A 3 (group) × 4 (saccade type: correct prosaccades and correct, erroneous, and corrective antisaccades) ANCOVA showed that RTs decreased quadratically with age, F1,98=7.94, P=.006, decreasing rapidly initially and reaching a plateau in late adolescence. There was an effect of saccade type, F1.9,190.0=211.54, P < .001: All pairwise comparisons were significant. There was also an interaction between the linear trend for age and saccade type, F1.9,190.0=12.10, P < .001. RTs of correct and corrective antisaccades, but not prosaccades or erroneous antisaccades, decreased linearly with age.
The diagnosis effect, F2,98=5.59, P=.005, and the diagnosis-by-saccade type interaction, F3.9,190.0=3.96, P=.005, were both significant. RTs of correct antisaccades were longer in the psychosis than in the control group, and RTs of corrective antisaccades were longer in the ADHD than in the control group. The RT difference between correct pro- and antisaccades did not differ among groups.
We obtained an effect of saccade type for latency to reach peak pupillary dilation, F2,200=39.65, P < .001. Latencies were shorter for correct prosaccades than for correct or erroneous antisaccades, which did not differ from each other.
Variability of RTs decreased linearly with age, F1,99=10.43, P=.002. There were effects of saccade type, F2.8,277.7=109.24, P < .001, and diagnosis, F2,99=20.40, P < .001, and an interaction between diagnosis and saccade type, F5.6,277.7=2.64, P=.019. RT variability was larger for corrective antisaccades than for the other saccade types and larger for prosaccades than for the other saccade types. Furthermore, saccadic RTs were more variable in the clinical groups than in the control group for prosaccades and correct and erroneous antisaccades but not for corrective antisaccades. (We examined whether greater variability of saccadic RTs in the clinical groups was attributable to the fact that they had fewer trials with saccades than the control groups. To test this hypothesis, we deleted the first and last 5 saccades in each category of saccades in the control group [correct pro- and antisaccades and erroneous and corrective antisaccades]. The ANCOVA on number of trials with saccades yielded a diagnosis effect, F2,104=3.30, P=.041 but no other significant result involving diagnosis. In this analysis, the control group had fewer trials than the ADHD group. Nevertheless, the ANCOVA on variability of saccadic RTs still yielded a diagnosis effect, F2,101=10.82, P < .001 [effects not involving diagnosis are not reported]. Post hoc analyses again indicated that RT variability was larger in the clinical groups than in controls and that the clinical groups did not differ from each other.)
To test if antisaccade errors or corrective saccades were related to fluctuations in attention or maintaining readiness to respond, we examined correlations between antisaccade errors and corrective saccades vs variability of RTs in the 2 conditions. As shown in table 5, correlations were weak to nonexistent for antisaccade errors. In contrast, we obtained much larger correlations in all 3 groups between greater variability of saccadic RTs and smaller likelihood of making corrective antisaccades.
In a large-scale study of healthy men, greater variability of antisaccade RTs was found to be correlated with lower IQ.70 To test if a similar association would be apparent in the current sample, we correlated variability of antisaccade RTs with IQ. Correlations between IQ and variability, with age partialed out, were not significant in the control, r42=.14, P=.369, or ADHD groups, r20=.39, P=.071. In the psychosis group, however, lower IQ was correlated with higher variability, r23=−.45, P=.023.
Because controlling for IQ risks reducing the variance due to the disorders themselves, we repeated the analyses without covarying IQ. Other than the interaction between IQ and diagnosis for detected saccades, the only ANCOVA or post hoc result involving diagnosis that differed was that the main effect of diagnosis on pupillary dilations no longer reached significance. Therefore, lack of significant diagnosis effects in the analyses above are not attributable to the effects of covarying IQ.
IQ was not correlated with variability of prosaccade RTs or RTs of pro- or antisaccades in any of the groups. We also found no correlations between IQ and other key variables (errors in the antisaccade task, absolute pupillary diameters, pupillary dilations, and latency to reach peak dilation in the pro- and antisaccade tasks) in any of the groups.
As many psychiatric medications have anticholinergic properties that can influence memory and absolute pupillary diameter, an anticholinergic equivalent measure was calculated for each patient based on anticholinergic receptor–binding affinity.71–73 In addition, each participant's current dose of antipsychotic medications was converted to a chlorpromazine equivalent, which provides an estimate of D2-blocking activity.74 Correlations between anticholinergic and chlorpromazine equivalents and performance variables did not yield any significant results that would survive a Bonferroni correction (all P’s > .117).
Finally, we repeated the analyses after excluding 3 participants in the psychosis group who were on clozapine. Only 2 of the ANCOVA and post hoc results reported above involving diagnosis changed: The interaction between IQ and diagnosis for detected saccades no longer reached significance, P=.051, and RTs of correct antisaccades no longer differed significantly between the psychosis and control groups, P=.078.
To our knowledge, this is the first published study in which pupillary dilations have been measured on pro- and antisaccade tasks. Therefore, we will first discuss the implications of pupillometric data for pro- and antisaccade tasks. Next, we discuss the clinical implications of all the findings.
We found larger dilations and pupillary diameters for anti- than for prosaccades. Similar coupling between larger diameters and dilations on harder tasks have been observed on other tasks.68,75 Furthermore, a higher rate of errors on the antisaccade task were strongly associated with smaller pupillary dilations on error trials in the control group. These results are consistent with the notion that antisaccades are more effortful or resource demanding than prosaccades and extend research on pupillary dilations to internally vs externally guided saccades.
As can be seen in figure 1, pupillary dilations for correct pro- and antisaccades showed an initial peak, followed by a gradual rise, especially in the prosaccade condition. The initial peak at 400–500 milliseconds in the prosaccade condition is likely to be related to the cognitive and motor processes involved in detecting the target and preparing and executing a visually guided saccade. (We are grateful to Stuart Steinhauer, PhD, and Eric Granholm, PhD, for their suggestions on the interpretation of these waveforms.) Richer et al76 also report that in simple RT tasks involving manual responses, peak pupillary dilations of similar magnitude occur about 500–600 milliseconcds after response onset. The greater magnitude and longer latency of the initial peak in the antisaccade condition likely reflect the additional effort and cognitive and motor processes involved in inhibiting the prepotent response and moving visual-spatial attention covertly from one side of the computer screen to the other.
The interpretation of the second rise in pupillary dilations is less clear. Participants were not given precise instructions for when to return to fixation, and pupillary dilations were not measured during the intertrial interval. There was also considerable variability among participants in the likelihood and timing of subsequent saccades during the trial; so, we could not conduct a systematic analysis of dilations in relation to these saccades. Therefore, we speculate that this second rise might be related to cognitive and motor processes involved in the recognition of having correctly performed the task, the decision about when to return to fixation, and execution of the saccade back to the fixation point.
Although erroneous antisaccades are sometimes viewed as being essentially the same as prosaccades,23 these results suggest otherwise. Saccadic RTs were shorter for erroneous antisaccades than for prosaccades. Pupillary dilations and latencies to reach peak dilation did not differ between correct and erroneous antisaccades; however, dilations were larger and latencies to reach peak dilation longer for erroneous than for correct prosaccades. These results strongly suggest that errors on the antisaccade task are qualitatively different from prosaccades, although both types of saccades superficially involve the same motor response directed at the same visual target. Variability of RTs was also larger for corrective antisaccades than for any other type of saccade, consistent with the hypothesis that corrective saccades may consist of both express saccades and more consciously corrected errors resulting in slower saccades.
As can be seen in the control data in figure 2, the pupillary waveform for erroneous antisaccades is clearly distinct from that for prosaccades and more similar to that for correct antisaccades in all groups. These waveforms are consistent with models that posit parallel activation of 2 competing processes engaged in a race to reach a threshold, one related to the prosaccade and the other to the antisaccade.24 The sustained elevation in the pupillary waveform of erroneous antisaccades after the initial peak, at least in the control and psychosis groups, might reflect the (conscious or subconscious) awareness of having made an error and preparation and execution of a corrective saccade.
These data are also noteworthy in that they indicate a clear dissociation between speed of processing and resource expenditure. RTs of erroneous antisaccades were shorter than RTs of prosaccades, yet pupillary dilation was larger on erroneous antisaccades.
Compared with controls, the psychosis—but not the ADHD—group made more errors on the antisaccade condition. These results extend the robust finding of increased antisaccade errors in adults with schizophrenia to youth-onset psychosis. The current results rule out 2 explanations for the higher error rates. First, these errors are unlikely to have been due to attentional fluctuations, as the correlations between antisaccade errors and variability of saccadic RTs were not significant in the psychosis group. Second, it has been suggested that elevated antisaccade error rates in clinical disorders may be due to abnormally high levels of preparatory activity in the motor system prior to saccade execution.77 The pupillary dilation data argue against this explanation for the psychosis group, who had smaller pupillary dilations during correct antisaccades than both the control and ADHD groups.
Instead, the pupillary data suggest that antisaccade errors in the psychosis group were accompanied by less phasic resource recruitment (ie, less cognitive effort). Phasic resource recruitment was also found to be impaired on an auditory digit span task in the same sample of participants.61 These results, which extend previous results of reduced pupillary dilations on other cognitive tasks in adults with schizophrenia to antisaccades in youth with psychosis, are consistent with behavioral studies indicating that individuals with schizophrenia allocate less effort to tasks than controls,78 fail to show increases in regional activation of the prefrontal cortex and striatum during antisaccade performance,25,79 and fail to show increased contingent negative variation amplitudes on anti- compared with prosaccades.80,81
It should be noted, however, that although smaller pupillary dilations on the antisaccade task were strongly correlated with higher error rates in the control group and although the correlation was also in the negative direction in the psychosis group, it was weak and did not approach significance. Thus, the higher error rates in the psychosis group were accompanied by, but not explained by, less effort. It is possible that greater effort translates more directly into better performance only when the neural substrates of antisaccades are intact.
Contrary to expectations, differences between the control and ADHD groups did not approach significance for antisaccade errors. The effect size of the control-ADHD difference for these measures was small (0.20), and a power analysis indicated that we would have needed 292 participants per group to achieve 80% power. The pupillary dilation data also provide compelling evidence that phasic recruitment of resources for both pro- and antisaccades was normal in ADHD. The effect sizes of the differences between the control and ADHD groups in peak pupillary dilation were very small, and as shown in figure 1, the initial rise was almost identical between groups. The discrepancy between our results and those of previous studies could be attributable to several factors. In our companion study with a smaller and older sample, eg, we found increased antisaccade error rates on a 32-trial version of the same task in the ADHD group compared with both age-matched and younger controls.28 The difference appears to be related to the performance of the control group: Error rates were very similar in the ADHD group between the current study (54%) and in the study of Karatekin28 (45%–50%), whereas average error rates in the age-matched control group were much lower (26%) in the study of Karatekin28 than in the current study (49%). The difference is likely to be due to the older age of the control sample in the study of Karatekin28 (M=180 mo) compared with the current sample (M=147 mo). In a study of adults with ADHD as well, error rates averaged 16% in the control group but 40% in the ADHD group.31 Task parameters and instructions could also contribute to discrepancies across studies. In 2 other studies with adults, for instance, accuracy rates were approximately 10% in the controls and 20% in the ADHD group.30,32 Thus, elevated error rates on the antisaccade task may not be as robust a phenomenon in ADHD as in schizophrenia, and their presence may depend, at least in part, on the age range of the participants and on task characteristics.
As can be seen in figure 1, however, the second dilation after the initial peak was larger in the ADHD than in the control group. We speculate that this difference might be related to greater uncertainty regarding when to return to fixation or about task performance or a greater frequency of saccades after the initial saccade in the ADHD compared with the control group.
Both clinical groups corrected fewer antisaccade errors than controls, and the psychosis group showed a trend toward fewer corrective saccades than the ADHD group. Of all the measures in this study, corrective saccades yielded the largest effect size for the control-psychosis difference and the second largest for the ADHD-control difference.
Despite the smaller sample size and number of trials in our companion study,28 we found that the ADHD group made fewer corrective saccades compared not only with age-matched controls but also younger children. It was suggested that the low rate of corrections in ADHD might have been due to difficulties in maintaining task instructions in mind. Deficiencies in working memory for task instructions have also been suggested as an explanation for the elevated rates of antisaccade errors in individuals with schizophrenia81–84 and ADHD34 and in healthy individuals.85,86 However, the current data are inconsistent with this hypothesis for both disorders. If the participants in the clinical groups had difficulty keeping task instructions in mind and were simply making prosaccades instead of antisaccades, we might have expected the parameters of their erroneous antisaccades to resemble those of prosaccades. This was clearly not the case. Instead, results suggest that the erroneous antisaccades of participants in the clinical groups differed from regular prosaccades just as the controls did. Thus, it seems likely that participants in both clinical groups were aware, at least at some level, that they were instructed to make antisaccades.
It was also suggested in our companion study that motivational factors may have played a role in the smaller proportion of corrective saccades in the ADHD group. However, the pupillary dilation data in the current study suggest that at least the ADHD group was putting in the same amount of effort into the initial saccades as controls, and both clinical groups were motivated enough to correct their errors at least 75% of the time on average. Nevertheless, we cannot rule out the possibility that the clinical groups may have been less motivated than controls to correct their errors.
However, there is stronger evidence that the failure to correct antisaccade errors was related to attentional fluctuations in both clinical groups. First, both groups had more variable saccadic RTs than controls. Indeed, of all the measures analyzed in this study, variability of antisaccade RTs yielded the largest effect size for the difference between the ADHD and control groups and the second largest effect size for the difference between the psychosis and control groups (table 3). Variability of manual RTs was also a prominent finding in the same sample on other tasks administered in our laboratory.61 Furthermore, greater RT variability was related to a lower likelihood of correcting errors. It is important to note that although the correlations were larger for variability of correct and erroneous antisaccades, variability of RTs to make simple visually guided saccades also predicted the probability of correcting inhibitory errors in the antisaccade condition but not the probability of errors themselves.
In healthy adults, more than half of erroneous saccades may go unrecognized, although the majority of these errors are corrected.14,87 There are also large interindividual differences in the degree to which errors are recognized.88,89 Different components of event-related potentials have been related to recognized vs unrecognized errors, although corrective saccades are often instituted whether errors are consciously recognized or not.15 In future research, it would be worthwhile to examine the degree to which individuals with psychosis or ADHD recognize their errors and whether their failure to make corrective saccades is associated with recognized or unrecognized errors.
Compared with controls, the psychosis group had longer antisaccade RTs. The ADHD group had elevated RTs for corrective antisaccades only. Importantly, however, the difference between correct pro- and antisaccades did not differ among groups, indicating that antisaccades RTs were not disproportionately longer in the clinical groups compared with the control group. There were no group differences in latency to reach peak dilation for pro- or antisaccades. Thus, the elevated saccadic RTs might be related to processes that occur closer to the response stage rather than processes that occur closer to the stimulus detection and evaluation stages.
Absolute pupillary diameters did not differ among groups in either condition. However, diameters were larger in the anti- than the prosaccade condition in the control but not in the clinical groups. These results suggest a deficit in both clinical groups in regulating tonic level of arousal to meet greater task demands.
It should be noted, however, that we did not find similar diagnosis-by-condition interactions in the same sample of participants on other tasks that included conditions of varying difficulty.61 Therefore, impairments in regulating tonic level of arousal do not appear to generalize across tasks in these 2 clinical groups.
Consistent with other developmental studies,29 error rates and RTs in the antisaccade task showed large improvements with age, whereas prosaccade RTs did not. There was a clear age-related improvement in correcting antisaccade errors, also replicating previous studies.17
Another cross-sectional study of pro- and antisaccades in ADHD also found no group-by-age interactions between ages 7 and 15 years for antisaccade errors34 but a more gradual decrease in antisaccade RTs and in express saccades in the prosaccade condition in the ADHD than in the control group between ages 7 and 10 years. In the current study, almost none of the analyses yielded age-by-diagnosis interactions, although there were large age effects on most of the variables. With the caveat that this was not a longitudinal study, results suggest that the groups did not differ in terms of the maturation of the cognitive processes assessed by the tasks in this study between 8 and 20 years of age.
The eye monitor had a relatively low temporal resolution of 60 Hz. However, this limitation should not have affected recording of saccadic RTs, as a 10° prosaccade should have a peak velocity of 200–400°/s and last 30–50 milliseconds, well above the resolution of the monitor.90–97 Furthermore, internally generated saccades, including antisaccades, have lower peak velocities and longer durations than visually guided saccades.16,95–100 Saccade duration and peak velocity do not show large developmental changes during adolescence.101,102 Individuals with ADHD show elevated saccade durations31,36 and reduced36 or normal peak velocity.103 Individuals with schizophrenia also generally show normal saccade durations and normal or reduced peak velocity23,25; increased peak velocity has also been observed.104
The data presented in figures 1 and and22 suggest that the pupil started to dilate again after the initial dilation, and we assume that the second dilation returned to baseline levels by the end of the intertrial period. However, this assumption needs to be tested empirically,105 and it is clearly necessary to have longer periods of recording during and between trials in future studies of pro- and antisaccade tasks.
The current study was part of a larger study that included 2 cognitive testing sessions for all and brain imaging for some participants. This procedure likely excluded severely impaired participants and chaotic, dysfunctional, or low–socioeconomic status (SES) families who were unable or unwilling to invest the necessary time to participate in the study, limiting generalizability of the results. Although the SES in the psychosis group was in the average range, it was significantly lower than that in the other groups.
We could not assess the effects of comorbid conditions in the clinical groups adequately due to small sample sizes. Results are limited to the combined subtype of ADHD. Additionally, we had few girls in the ADHD group, as it was difficult to find girls who met criteria for the combined subtype.
Because there was no Bonferroni correction across terms in the ANCOVAs, some of the results, especially higher order interactions, close to a P level of .05 may have been spurious.
Participants with psychosis were on a variety of medications. Although correlational analyses did not show any relation between anticholinergic and chlorpromazine equivalents and performance measures, medication effects in the psychosis group cannot be ruled out. However, abnormalities in pupillary responsivity were observed in patients with schizophrenia in the early 1900s, before the introduction of current medications.42 Studies examining medication effects on pro- and antisaccades in schizophrenia have not reported significant effects of the medications in the current study on performance80,106–108; impairments on the antisaccade task have been reported in neuroleptic-naive first-episode patients with schizophrenia109–112 and in unmedicated relatives of individuals with schizophrenia,113,114 and antisaccade errors have been shown to be positively correlated with schizotypal or psychosis proneness traits in the normal population.115,116 In addition, studies of adults with schizophrenia show that risperidone, an atypical neuroleptic, is either associated with reduced antisaccade error rates and antisaccade RTs109,117 or has no effect on antisaccade errors and RTs.110 However, medications may be associated with other measures, such as RT of visually guided saccades.110,118
Although not strictly a limitation of the study, it should be noted that the experimenters provided structure for the participants as necessary in order for them to complete the tasks. In many cases, the experimenter sat behind the participant, but it was sometimes essential to stand next to the participants throughout a condition to encourage them to continue. Although participants were discouraged from talking during the task, they differed in how much they spontaneously talked to the experimenter or to themselves. There were also differences in how many reminders participants were given to follow the instructions, particularly on the antisaccade condition. We did not quantify differences on these variables. Had testing conditions been identical across groups and had the experimenters provided less structure, we would have had more missing/invalid data but we might also have had larger group differences.
The psychosis, but not the ADHD, group had elevated antisaccade error rates. Furthermore, these inhibitory failures in the psychosis group were accompanied by a lower level of momentary cognitive effort (as indexed by pupillary dilations). These findings extend previous research pointing to resource limitations in adults with schizophrenia45,46,119 to youth-onset psychosis. In contrast, the data from error trials do not support the hypothesis that individuals with schizophrenia make more antisaccade errors because of difficulties with maintaining an internal representation of the task instructions.120
Impairments in phasic regulation of resources could be related to impairments in either the phasic inhibition of the parasympathetic pathway controlling pupillary constriction or phasic activation of the sympathetic pathway controlling pupillary dilation. These pathways receive direct and indirect inputs from the reticular activating system, the frontal cortex, the basal ganglia, the hypothalamus, and the thalamus. The smaller dilations observed in the current study could stem from impairments in these regions or in the phasic interactions among them, perhaps in the regulation of the reticular system by cortical regions.48 Thus, these pupillometry data are consistent with the theory that schizophrenia is fundamentally a disorder of functional connectivity in cortical-subcortical neural networks.121
With respect to ADHD, results challenge theories and models that posit inhibition as a core cognitive impairment.122 To the extent that pupillary dilations reflect voluntary allocation of effort, results are also inconsistent with the proposal that inhibitory failures in ADHD reflect a motivational deficit.123
We also observed deficits that were not specific to either disorder. Indeed, the largest differences between the control and clinical groups were found not in the expected indices of inhibition, but in the probability of correcting inhibitory errors and in variability of antisaccade RTs, which were correlated with each other. Thus, results point to impairments in both clinical groups in sustaining attention on a trial-by-trial basis, resulting in deficits in self-monitoring. Together with other studies in ADHD (Mostofsky et al35 and Munoz et al36), these results extend the common finding of variable manual RTs in both disorders to the oculomotor domain and demonstrate that increased RT variability is not specific to either disorder. Such attentional fluctuations, which may prevent individuals with psychosis or ADHD from consciously detecting and correcting the erroneous consequences of their voluntary actions, would have significant implications for performance on a wide range of laboratory and everyday tasks.
These findings further suggest that improving top-down control of attention over time should be a target of cognitive remediation efforts in both ADHD and psychosis. It is an empirical question what kinds of cognitive interventions would be most effective for attentional fluctuations and whether the same strategy would be equally effective in both disorders given that the same impairments occur in the context of different cognitive profiles. Phasic resource recruitment has not been included as a component of neurocognitive training programs in psychosis; it would be informative to determine whether it is responsive to interventions.
National Institute of Mental Health (1RO3-MH063150 to C.K., K08-MH068540 to T.W.); National Alliance for Research on Schizophrenia and Depression (to C.K., T.W.); Essel Foundation (to T.W.); University of Minnesota Center for Neurobehavioral Development (to C.K., T.W.).
We thank the families and teachers for participating in the study; Afshan Anjum, Angie Guimaraes, Bonnie Houg, Cacy Miranda, Kathryn McGraw-Schuchman, and Marie Gabrielle Reed for helping with diagnostic assessments; Clay Collins, Nicholas Davenport, Anita Fuglestad, David Marcus, and Marcus Schmidt for helping with data collection and analyses; and research assistants for helping with data entry and organization.