Cognitive Control Under Contingencies in Anxious and Depressed Adolescents: An Antisaccade Task Sandra Jazbec, Erin McClure, Michael Hardin, Daniel S. Pine, and Monique Ernst Background: Emotion-related perturbations in cognitive control characterize adult mood and anxiety disorders. Fewer data are available to confirm such deficits in youth. Studies of cognitive control and error processing can provide an ideal template to examine these perturbations. Antisaccade paradigms are particularly well suited for this endeavor because they provide exquisite behavioral measures of modulation of response errors. Methods: A new monetary reward antisaccade task was used with 28 healthy, 11 anxious, and 12 depressed adolescents. Performance accuracy, saccade latency, and peak velocity of incorrect responses were analyzed. Results: Performance accuracy across all groups was improved by incentives (obtain reward, avoid punishment). However, modulation of saccade errors by incentives differed by groups. In incentive trials relative to neutral trials, inhibitory efficiency (saccade latency) was enhanced in healthy, unaffected in depressed, and diminished in anxious adolescents. Modulation of errant actions (saccade peak velocity) was improved in the healthy group and unchanged in both the anxious and depressed groups. Conclusions: These findings provide grounds for testing hypotheses related to the impact of motivation deficits and emotional interference on directed action in adolescents with mood and anxiety disorders. Furthermore, neural mechanisms can now be examined by using this task paired with functional neuroimaging.

Key Words: Anticipation, eye movement, latency, motivation, peak velocity, punishment, reward

A

number of current theories view mood and anxiety disorders as conditions that result from developmental perturbations in information processing (Gotlib et al 2004; Pine et al 1998, 2000), and in emotion regulation (Davidson et al 2002). These theories emphasize the role of perturbed attention allocation, particularly as a function of affective context, in mood and anxiety disorders (Beuke et al 2003; Dalgleish et al 2003; Williams et al 1996). Furthermore, such attention– emotion perturbations are expressed in distinct patterns of motivated behaviors. A global lack of motivation characterizes depressive disorders, and exacerbated avoidant behavior with enhanced emotional interference characterizes anxiety disorders (Austin et al 2001; Ehrenreich and Gross 2002). Although such theories are generally well supported in studies of adults, virtually no research in adolescents employs basic neuroscience measures to explore associations between mood or anxiety disorders and either perturbed attention control or contingency-related information processing (Austin et al 2001). This limitation hinders efforts to move developmental models beyond simple descriptive formulations into neuroscience-based theories. Emotion-related attention bias in adolescent mood and anxiety disorders may manifest as perturbed control of saccadic eye movement. Because saccades index attention allocation (Brockmole et al 2002; Godijn and Theeuwes 2003; Munoz et al 2004), paradigms that examine motivational influences on saccadic control might identify aberrant attention– emotion interactions. Recent studies have begun to examine attention– emotion interactions in both adult and adolescent disorders. In adults, rela-

From the Section of Developmental and Affective Neuroscience, National Institute of Mental Health, National Institutes of Health, Bethesda, Maryland. Address reprint requests to Monique Ernst, M.D., Ph.D., Section of Developmental and Affective Neuroscience, Mood and Anxiety Disorders Program, NIMH/NIH/HHS, 15K North Drive, Bethesda, MD 20892; E-mail: [email protected]. Received January 4, 2005; revised March 14, 2005; accepted April 8, 2005.

0006-3223/05/$30.00 doi:10.1016/j.biopsych.2005.04.010

tively consistent attention biases are found in mood and anxiety disorders (Gotlib et al 2004). Some forms of emotion-related attention bias may occur in both anxiety and depressive disorders; other forms may represent disorder-specific deficits (Dalgleish et al 2003). In adolescents, however, the few available studies demonstrate less consistent associations than those found in adults, possibly because of the use of attention allocation measures that assess global dimensions of behavior (e.g., accuracy or reaction time of button press). These measures may lack sensitivity and contribute to heterogeneity of findings (Ehrenreich and Fischer 2002). Saccadic eye movement tasks provide a precise assessment of some aspects of top-down cognitive control processes that can influence attention allocation (Munoz et al 2004; Ridderinkhof et al 2004). Of particular interest are the measures indexing processes involved during the preparation for saccadic action and those during the execution of action. Because incorrect responses and saccade latencies during antisaccade tasks reflect the ability to inhibit prepotent responses (Munoz and Everling 2004), these measures may be particularly susceptible to modulation by motivationally salient stimuli, and particularly relevant to attention– emotion related deficits in mood and anxiety disorders. The neural circuitry underlying motivational influences on eye movement execution has been exquisitely delineated in nonhuman primates (Ikeda and Hikosaka 2003; Kobayashi et al 2002; Takikawa et al 2002; Watanabe et al 2003a, 2003b). This circuitry shows commonalities with the structures implicated in cognitive control processes, including response inhibition and error/conflict monitoring (Holroyd et al 2002; Tucker et al 2003), depression (Drevets 2001), and anxiety (Kent and Rauch 2003). Thus, assessments of motivational influences during saccadic errors may index perturbed cognitive control processes observed in adolescent mood and anxiety disorders. The antisaccade task is one of the most frequently used paradigms in developmental research on eye-movement control (Luna et al 2004a, 2004b). Antisaccades are rapid voluntary eye movements that are directed towards the mirror position of a cue presented in the peripheral visual field (Everling and Fischer 1998; Munoz and Everling 2004). To perform a correct antisaccade, the reflexive urge to look at the target (i.e., prosaccade) needs to be inhibited and a voluntary saccade has to be BIOL PSYCHIATRY 2005;58:632– 639 © 2005 Society of Biological Psychiatry

S. Jazbec et al programmed in absence of visual input. Not surprisingly, erroneously triggered prosaccades are common in this paradigm (Munoz et al 1998; Olincy et al 1997). The three most common performance parameters used in this task include accuracy, saccade latency, and saccade peak velocity (Leigh and Zee 1999; Munoz and Everling 2004). Accuracy provides a relatively global index of quality of performance, whereas latency and peak velocity are more precise measures of cognitive control processes that map to specific neural function. For example, prolonged saccade latency during antisaccades (compared with prosaccades) reflects decreased saccade neuron activity in the frontal eye fields and superior colliculus (Everling et al 1998; Munoz et al 2004). Shorter latencies during incorrect antisaccades are normal and reflect an inability to inhibit saccade neurons (Munoz et al 2004). Less efficient cognitive control predicts that the capacity to inhibit an erroneous response requires more time (i.e., longer latencies). Peak velocity, on the other hand, may assess motor regulation once execution of a saccade is initiated. In the context of an already initiated errant saccade, lower peak velocity may represent an effort to reduce the error as it is being executed. In this instance, reduced peak velocity reflects greater control exerted over the inappropriate saccadic action. The current work examines the degree to which incentives modulate processes engaged during saccadic errors in adolescents with mood and anxiety disorders. Specifically, we developed a novel monetary reward saccade task to assess attention–reward interactions. Unlike other tasks that tap attention– emotion interaction, this task assesses relatively subtle aspects of attention control under neutral, negative, and positivevalenced conditions: 1) failure to obtain an expected monetary gain (“reward” errors), 2) failure to avoid monetary loss (“punishment” errors), and 3) no monetary consequences (“neutral” errors). The study tests three sets of hypotheses. 1) Overall, accuracy rate of correct responses will be improved by contingencies; however, patient groups will show less efficient facilitation of performance by incentives, because of motivation deficits in depression and emotional interference in anxiety. 2) Cognitive control, assessed by saccade latency and peak velocity, will be enhanced by incentives in healthy subjects, as evidenced by decreased latency and decreased peak velocity of errant saccades; however, depressed and anxious adolescents will show lesser degrees of incentive-related minimization of an errant action. 3) Anxiety will affect performance on punishment trials more strongly than on reward trials because of negative bias in this disorder (Dalgleish et al 2003). Depression will influence response to both reward and punishment trials because motivation deficits affect responses to both positive and negative contingencies (Naranjo et al 2001).

Methods and Materials Participants The sample consisted of 28 healthy adolescents, 11 adolescents with anxiety disorders, and 12 adolescents with major depressive disorder (MDD). The Institutional Review Board of the National Institute of Mental Health approved the study. Parents gave written informed consent, and adolescents gave written assent prior to participation, after the study was fully explained and all questions answered. Patients were recruited when they sought treatment for a mood or anxiety disorder.

BIOL PSYCHIATRY 2005;58:632– 639 633 Healthy subjects were recruited through advertisements and contacts with medical organizations. Inclusion criteria for all subjects were age between 9 and 17 years, absence of acute or chronic medical problems, absence of any treatment with psychotropic medications for 1 month (2 months for fluoxetine), absence of severe trauma history or posttraumatic stress disorder (PTSD), absence of obsessive– compulsive disorder (OCD), chronic tic disorder, mania, conduct or oppositional defiant disorder, substance abuse, pervasive developmental disorder, or attention-deficit/hyperactivity disorder (ADHD) of sufficient severity to require immediate treatment. Other exclusion criteria comprised mental retardation (IQ ⬍ 70), use of any medication, and pregnancy. All participants were tested for IQ prior to entering the study with the Wechsler Abbreviated Scale of Intelligence (WASI; Wechsler 1999). Patients in the MDD group met DSM-IV criteria for current MDD and were required to exhibit elevated symptoms on the Child Depression Rating Scale (CDRS ⱖ 39). Patients in the anxiety disorder group met criteria for lifetime generalized anxiety disorder (GAD) and were required to show elevated symptoms on the Pediatric Anxiety Rating Scale (PARS ⬎ 10). Immediately following assessment and testing (including eye movement tracking), patients were provided treatment. Healthy adolescents met criteria for no current or past psychiatric disorders. All participants were evaluated through semistructured psychiatric interviews using the Kiddie Schedule for Affective Disorders and Schizophrenia for School-Age Children (K-SADS-PL). These evaluations were performed by experienced clinicians who each had demonstrated acceptable inter-rater reliability (␬ ⬎ .75) for all relevant diagnoses (Kaufman et al 1997). Procedures Recordings were obtained in a room lit by standard overhead fluorescent lights. Following initial calibration of eye position, eye movements were measured with high-resolution infrared oculography (Applied Science Laboratories [ASL] Model 504; Bedford, Massachusetts). Calibration was repeated between runs as needed. Before performing the task, subjects were thoroughly trained to prevent any learning effect. They also were debriefed after the completion of the task. Task The task measured rapid reflexive and voluntary eye movements in three contingency contexts: monetary gain (reward condition), monetary loss (punishment condition), and no incentive (neutral condition). It comprised three phases: 1) the initial cue phase (1250 –170 msec), which informed the subject about the type of trial (prosaccade or antisaccade; reward, punishment, or neutral); 2) the target phase or saccade phase (1850 msec); 3) and the feedback phase (1000 msec; Figure 1). Each trial started with one of six cues displayed at the center of a black computer screen, and subtending approximately .5° visual angle. The cues included a plus sign (⫹), a minus sign (–), or a small circle (䡩), presented in either white or gray. White cues signaled a prosaccade (i.e., an eye movement towards the target), and gray cues signaled an antisaccade (i.e., an eye movement to the mirror position of the target). The shape of the cue indicated the valence of the trial: a plus sign meant a $1 monetary gain for a correct eye movement or no gain for an incorrect eye movement (reward condition); a minus sign meant a $1 monetary loss for an incorrect eye movement, or no loss for www.sobp.org/journal

634 BIOL PSYCHIATRY 2005;58:632– 639

S. Jazbec et al Eye Movement Recording Eye movements were measured with an ASL Model 504 eye tracker with remote pan and tilt optics, autofocusing lens, and with magnetic head tracker. The spatial accuracy of the eye tracker is .25° visual angle. Sampling rate is 60 Hz. Use of a magnetic head tracking and an autofocusing lens minimized the possibility of artifacts due to head movements. Participants were instructed to remain still, and a chinrest was employed when necessary. Differences in eye-screen distance across subjects were corrected for in the offline analysis of the raw data. The distance from the eye to the screen was on average 26.35 ⫾ 1.96 inches for healthy subjects and 25.72 ⫾ 1.91 for patients.

Figure 1. Paradigm of the Reward Saccade Task: A cue (1250 to 1750 msec duration) is presented at the onset of each trial. The cue indicates the type of trial (gray for antisaccade and white for prosaccade) and the incentive condition of the trial (‘o’ for neutral, ‘⫹’ for gain, and ‘⫺‘ for loss). As the cue disappears, a target appears on the right or left side of the screen (1850 msec duration), until the feedback appears for 1000 msec. As the feedback disappears, the next trial starts with appearance of the central fixation cue. The number of the trials for each condition is presented below. Antisaccade

⫹ ⫺ o Total

Prosaccade

Right

Left

Right

Left

Total

12 12 12 36

12 12 12 36

12 12 12 36

12 12 12 36

48 48 48 144

.

a correct eye movement (punishment condition); and a circle meant the absence of monetary incentive (neutral condition). After a variable period of 1250 –1750 msec, the cue was replaced by a white target stimulus that appeared laterally on the screen. The target, an asterisk subtending .5° visual angle, appeared at approximately 6.15° eccentricity on the horizontal meridian either to the left or the right of the centrally located cue position. The target duration was 1850 msec. To succeed on a trial, subjects had to fixate for at least 100 ms an area of 60 pixels radius around the correct location. Fixation also had to occur within 500 msec after target appearance. Subjects were asked to maintain fixation until they received feedback. Feedback (1000 msec) was presented 1850 msec after target onset and subtended approximately 1.8° visual angle. Feedback consisted of dollar amounts (⫹$1, ⫺$1, $0). The feedback appeared at the location where the subject was supposed to have gazed, replacing the target in the prosaccade trials, or appearing in the mirror location of the target in the antisaccade trials. Subjects were asked to keep their eyes on the appropriate site until they saw the feedback cue. The feedback cue was green when the gaze was correct and red when the gaze was incorrect. The task consisted of three runs of 4 min duration each. Each run comprised 48 trials, with four trials per side (right, left) and condition (antisaccade-reward, antisaccade-punishment, antisaccade-neutral, prosaccade-reward, prosaccade-punishment, and prosaccade-neutral). The task included a total of 144 trials (24 trials per condition). Subjects started with $0 and could win up to $48 per run. Control adolescents won on average $25.8 ⫾ $10.9, anxious patients $23.5 ⫾ $13.5, and depressed patients $26.3 ⫾ $12.2. Participants were told that they would receive the dollar amount they won and were sent a check at the completion of the study. www.sobp.org/journal

Eye Movement Analysis The raw data were analyzed offline with software provided by ASL (EYENAL). This program calculates fixations (gazes) based on an algorithm that takes into account the distance of the eye to the screen for each subject. According to this fixation algorithm, a fixation starts when the standard deviation of the x and y coordinates of six consecutive samples (corresponding to 100 msec duration) is below .5° visual angle. A saccade was defined as an eye movement between two fixations. The saccadic measures used in this study were latency and peak velocity. The latency is defined as the time period elapsed between the onset of the target and the onset of the first subsequent saccade. The peak velocity was calculated as the saccade amplitude in degrees visual angle divided by half the saccade duration in sec (Carpenter 1988). Participants can make three kinds of saccades: anticipatory (latency ⬍ 80 msec), direct response (80 msec ⬍ latency ⬍ 700 msec), and delayed response (latency ⬎ 700 msec; Fischer and Weber 1992; Klein et al 2003). Because more than 90% of the saccades were direct saccades, only direct saccades were examined in this study. A correct response was defined as a saccade that occurred within 80 msec and 700 msec after target onset and ended up with the pupil projecting on a point within a circle of a 60 pixels radius around the correct point (mirror location of the target on a horizontal line in antisaccades). Because this study focused on behavior during errors, the analysis sampled parameters of direction errors of antisaccades, that is, unwanted reflexive prosaccades. In addition to accuracy (number of incorrect antisaccades), two independent parameters were analyzed: latency and peak velocity. We did not include saccade amplitude and duration because these variables are highly correlated with peak velocity, given the stereotyped pattern of saccades (Fischer et al 1992). Repeated-measures analyses of variance used saccade performance variables as dependent measures with diagnostic group (control vs. anxious vs. MDD) as the between-subjects factor, and accuracy (correct vs. error) and contingency (punishment vs. reward vs. neutral) as the within-subjects factors. We treated subjects with MDD alone or MDD comorbid with anxiety as a single group for two reasons. First, prior studies of attentional dysfunction in adolescent MDD find that individuals diagnosed with MDD alone or MDD with anxiety perform similarly, but that they differ from adolescents with anxiety disorders alone (Taghavi et al 1999). Second, longitudinal and family-based studies suggest that adolescents frequently present with anxiety in the absence of MDD but that adolescent MDD virtually always presents with anxiety either concurrently or at some other point during development (Costello et al 2002). Finally, because groups were comparable in age, IQ, and gender,

BIOL PSYCHIATRY 2005;58:632– 639 635

S. Jazbec et al Table 1. Diagnoses in the Patient Groups Group (Subject No.) Anxious (1) Anxious (2) Anxious (3) Anxious (4) Anxious (5) Anxious (6) Anxious (7) Anxious (8) Anxious (9) Anxious (10) Anxious (11) Depressed (1) Depressed (2) Depressed (3)

GAD y y y y y y y y ya y y

SepAD

y y y y y y y

SocPh

SpecPh

y y y y y y

y

Only three depressed children are included in this table because no comorbid diagnoses were present in the other nine depressed patients. GAD, generalized anxiety disorder; SepAD, separation anxiety disorder; SocPh, social phobia; SpecPh, specific phobias. ya Lifetime history of GAD but not currently meeting severity criteria for GAD.

these variables were not included in the analysis. Significant main effects or interactions were decomposed using post hoc analyses. All conclusions are based on two-tailed tests with an a priori ␣ ⫽ .05. For completeness, correlations between performance scores and ratings of severity of symptoms (CDRS and PARS) were conducted in the combined patient sample and in each patient group separately. No significant correlations were found. Finally, a debriefing questionnaire was completed by all participants after the eye tracking study. We collected data on how difficult the task seemed to subjects and how well subjects could distinguish between the gray and white cue (4-point rating scales: 1 not at all, 4 extremely). Subjects rated the question “How difficult was the task?” between “a little and somewhat difficult” with a mean rating of 1.8 ⫾ .7 for the healthy group, 2.0 ⫾ .7 for the anxious group, and 1.8 ⫾ .6 for the depressed group. This subjective rating of difficulty was not correlated with age (r ⫽ –.12, n ⫽ 46, p ⫽ .4). In response to the question “How well could you make the difference between the gray and white cue?” the three groups scored in the “very well” range: 3.4 ⫾ .7 for the healthy group, 3.2 ⫾ .7 for the anxious group, and 2.9 ⫾ .9 for the depressed group. This subjective rating was not correlated with age (r ⫽ –.17 p ⫽ .3). Therefore, effects of difficulty did not differ among groups and could not account for the group differences in task performance.

Results Sample The sample included 28 healthy adolescents (13 boys, 15 girls; mean ⫾ SD: age 13.0 ⫾ 2.5; IQ 107.3 ⫾ 13.3), 12 depressed adolescents (4 boys, 8 girls; mean ⫾ SD: age 13.9 ⫾ 2.5; IQ 109.3 ⫾ 18.3), and 11 anxious adolescents (8 boys, 3 girls; mean ⫾ SD: age 12.2 ⫾ 1.6; IQ 112.7 ⫾ 10.9). As noted above, groups did not differ significantly on gender distribution, IQ, or age. All anxiety patients met lifetime DSM-IV criteria for GAD, and 10 of 11 met criteria for ongoing GAD (one did not meet criteria

for ongoing GAD but continued to meet criteria for separation anxiety disorder and social phobia). In addition, comorbid anxiety disorders in the anxiety group were common (see Table 1). Among participants with MDD, three had a comorbid anxiety disorder (see Table 1). Three anxious patients also met criteria for ADHD. However, the symptoms of ADHD were less severe than their anxiety symptoms. Performance on Antisaccade Trials: Incorrect Responses All means and standard errors are presented in Table 2. For completeness, data on correct responses are also included in this table, although analyses examine between-group differences only during saccadic errors. Performance scores during the neutral condition did not differ among groups, reflecting similar baseline performance in healthy, anxious, and MDD adolescents. Accuracy. The total number of recorded antisaccade responses per condition ranged between 21.7 and 22.8 for all groups. Thus, an average of two antisaccade responses were not detected by the eye-tracker in each group (total number of trials per condition ⫽ 24). This loss of data was due to blinking or to the eye camera losing the pupillary signal. The mean number of incorrect responses per condition ranged between 5.9 and 9.4 across the whole sample (Figure 2). For all three groups, as hypothesized, the number of incorrect responses was lower in the reward and punishment trials than in the neutral trials [Contingency: F (2,88) ⫽ 6.54; p ⫽ .002], reflecting better performance under reinforcement conditions (reward or punishment). However, no main group effect or group-by-contingency interaction emerged. Thus, incentive-related changes in accuracy did not differ significantly among groups. Table 2. Mean (SE) of Number of Incorrect and Correct Antisaccades, Latency to Saccades, and Peak Velocity by Group and Condition Healthy (n ⫽ 28) Punishment Reward Neutral Punishment Reward Neutral Punishment Reward Neutral Punishment Reward Neutral Punishment Reward Neutral Punishment Reward Neutral

Anxious (n ⫽ 11)

Depressed (n ⫽ 12)

Number of Incorrect Antisaccades 5.88 (.78) 7.73 (1.56) 5.89 (.71) 7.64 (1.63) 8.89 (.94) 9.36 (1.64) Number Correct Antisaccades 15.54 (.96) 13.55 (1.26) 15.04 (.90) 13.36 (1.22) 11.82 (1.11) 11.82 (1.57) Latency to Incorrect Antisaccades 184.11 (11.44) 223.72 (14.66) 185.60 (10.58) 213.73 (11.65) 208.21 (9.52) 191.41 (7.94) Latency to Correct Antisaccades 322.14 (11.24) 336.88 (15.10) 316.66 (8.68) 317.91 (14.73) 318.02 (9.95) 326.34 (12.48) Peak Velocity of Incorrect Antisaccades 280.61 (14.91) 323.39 (7.31) 273.73 (15.06) 325.31 (8.04) 313.51 (9.79) 326.76 (6.75) Peak Velocity of Correct Antisaccades 355.32 (8.21) 370.78 (15.40) 358.97 (10.02) 372.89 (9.89) 347.24 (10.16) 369.77 (17.11)

7.42 (.97) 8.00 (.78) 8.50 (1.60) 15.73 (1.21) 14.83 (1.49) 13.83 (1.49) 184.24 (17.05) 198.63 (12.45) 191.71 (16.61) 305.77 (16.83) 291.09 (17.64) 304.68 (20.06) 316.21 (20.82) 342.98 (10.13) 314.75 (16.79) 384.71 (10.85) 359.61 (19.08) 353.65 (21.60)

www.sobp.org/journal

636 BIOL PSYCHIATRY 2005;58:632– 639

S. Jazbec et al

Figure 2. Mean (SE) number of incorrect antisaccades (unwanted reflexive prosaccades) in healthy, anxious, and depressed groups, by incentive condition

Latency. In the case of errant saccades, longer latencies reflect the inability to perform correctly despite longer periods of preparation (Figure 3). As hypothesized, contingencies had distinct effects across groups for the latency of incorrect responses [group-by-contingency F (4,84) ⫽ 3.18; p ⫽ .017]. This interaction was decomposed by examining the effects of contingencies on latency separately in each of the three groups. For the healthy adolescents, contingencies were associated with expected reductions in latency [F (2,22) ⫽ 3.41; p ⫽ .05]. Similar effects emerged for both rewards and punishments relative to neutral contingencies, with no difference between the two incentive conditions; however, this effect did not emerge in adolescents with MDD, in whom there was no effect of either contingency on latency [F (2,9) ⫽ .72; p ⫽ .52]. For the anxious group, in contrast, incentives were associated with an increase in latency, particularly for punished trials [F (2,24) ⫽ 5.6; p ⫽ .027]. Thus, in each of the three groups, latency on incorrect trials exhibited a distinct

Figure 3. Mean (SE) latency of incorrect antisaccades (unwanted reflexive prosaccades) in healthy, anxious, and depressed groups, by incentive condition

www.sobp.org/journal

pattern of modulation related to the anticipation of positive or negative contingencies (see Figure 3). Peak Velocity. Peak velocity during the execution of a saccade error quantifies the engagement of compensatory mechanisms. The lower the velocity, the more inhibited the saccade. Thus, reduced velocity reflects the greater engagement of compensatory influences during the execution of an incorrect action. As with the latency variable, contingencies exerted distinct effects across patient groups for peak velocity on incorrect trials [group-by-contingency F (4,84) ⫽ 3.27; p ⫽ .015]. In healthy subjects, peak velocity was reduced by both reward and punishment incentives relative to neutral trials [F (2,46) ⫽ 4.96; p ⫽ .011]. This reflects the expected greater regulation in trials associated with potential monetary gain or loss. Anxious patients failed to show any alteration in peak velocity in the presence of incentives [F (2,20) ⫽ .11; p ⫽ .84]. Similarly to the anxious group, the MDD group failed to show any incentive-related alteration in peak velocity [F (2,20) ⫽ .89;

S. Jazbec et al p ⫽ .19], although the MDD group tended to exhibit an increase in peak velocity on reward trials. No differences emerged between the GAD and MDD groups.

Discussion This study presents results from a novel saccadic control task. Overall, the results demonstrate that attention-related processes engaged during errant execution of a saccade are sensitive to experimentally manipulated reward contingencies. Moreover, results point to perturbations in the degree to which these contingencies affect cognitive control processes in adolescents with anxiety or MDD, relative to healthy adolescents. The three hypotheses tested in this study were partly supported by the findings. 1) Overall, accuracy was improved by contingencies: fewer errors occurred during reward and punishment trials than during no-incentive trials. Accuracy did not differ among healthy, anxious, and MDD adolescents, however (Figure 2). 2) Cognitive control, indexed by saccade latency and peak velocity of errant antisaccades, was facilitated by contingencies in healthy adolescents: shorter latency and lower peak velocity occur in reward and punishment trials relative to noincentive trials. Anxious and depressed adolescents showed either no improvement or worsening of these parameters with contingencies. 3) The distinct effects of contingencies on performance as a function of positive, negative, and neutral conditions was different in anxious and depressed adolescents: anxious adolescents were most deviant in the negative condition, whereas depressed adolescents were abnormal in both positive and negative conditions. Depression was associated with a similar pattern of deficient influence of incentives on both saccade preparation and control of saccade execution, that is, absence of improved performance under both positive and negative incentives. In contrast, anxiety was associated with a different pattern of perturbed influence of incentives on saccade preparation and saccade execution, that is, worsening of efficiency of inhibition (saccade preparation) under positive and negative incentives, particularly negative incentives, and absence of improvement in minimizing error execution (saccade execution).

BIOL PSYCHIATRY 2005;58:632– 639 637 sence of group difference in accuracy suggests that this variable provides a global measure of performance, relatively insensitive to mood and anxiety disorders; however, this negative finding, as with all negative findings in this study, should be interpreted with caution given the small number of patients. Thus, a global performance deficit on this task related to mood and anxiety disorders cannot be ruled out. The measures of saccade latency and peak velocity during error execution probe specific processes that occur either before or during the engagement of an errant saccadic behavior. Latency is the period of time elapsed between the appearance of a lateral target and the initiation of a motor response toward this target (prosaccade) or away from this target (antisaccade). The antisaccade latency represents the period of time necessary to inhibit a reflexive saccade toward a suddenly appearing target and to initiate a saccade away from this target (Leigh and Zee 1999). Therefore, longer latencies indicate that subjects have more time to inhibit an errant target. Longer latencies of incorrect saccades are associated with a weaker capacity to inhibit a reflexive saccade: subjects still fail to inhibit unwanted reflexive saccades despite longer latencies. Peak velocity, a measure of the magnitude of a saccade, indexes control of motor execution. After an errant saccade is released, the reduction of the magnitude of this saccade reflects the capacity to reduce error after the movement is initiated (Leigh and Zee 1999): reduced peak velocity of incorrect saccades implies greater capacity to minimize error. Thus, these measures of capacity for inhibiting (latency) or minimizing (peak velocity) incorrect actions permit probing of the efficiency of cognitive control. Findings from the present study suggest that processes controlling both the preparation and execution of eye movements are modulated by incentives differently in youth with MDD and with GAD, relative to healthy adolescents. Latency: Efficiency of Inhibitory Preparatory Control Healthy adolescents showed enhanced performance (shorter latencies) under both positive and negative incentive conditions relative to nonincentive condition. This finding corroborates facilitation of behavior under contingencies (Wasserman et al 1997).

Figure 4. Mean (SE) peak velocity of incorrect antisaccades (unwanted reflexive prosaccades) in healthy, anxious, and depressed groups, by incentive condition

www.sobp.org/journal

638 BIOL PSYCHIATRY 2005;58:632– 639 In contrast, anxious adolescents responded with increased latency to the presence of incentives compared with the neutral condition in errant saccades. In other words, longer preparatory periods were not sufficient to inhibit errors. The punishment condition (potential monetary loss) was the most affected condition (longest latencies) in anxious adolescents (see Figure 3). This suggests that, whereas both positive and negative contingencies diminished efficiency of inhibiting a reflexive prosaccade in anxious adolescents, negative contingencies were more disruptive. This finding is consistent with the negative attention bias reported in this population (Ehrenreich and Gross 2002). We propose that the presence of contingencies, positive or negative, creates an interference effect. The possibility of reward (or failure to obtain reward) or of punishment may induce a state in which attentional resources are partly diverted away from cognitive control processes towards emotion based processes. This attention diversion would result in a reduction of the attentional resources allocated to inhibit a prepotent action (prosaccade). Deficits in cognitive control could reflect dysfunction of the anterior cingulate (Brown and Braver 2005; Carter et al 1998; Holroyd et al 2004) and hyperactivity in other limbic and paralimbic areas (Cannistraro and Rauch 2003). Such interpretation could be tested using functional neuroimaging paired with the reward saccade task. When contrasting brain activity during errant saccades in a reward or punishment condition with that during a nonincentive condition, we would expect reduced activation in the anterior cingulate, a key structure in cognitive control (Brown and Braver 2005; Carter et al 1998; Holroyd et al 2004) and enhanced activation in paralimbic and limbic structures such as ventral prefrontal cortex and amygdala in anxious adolescents compared to healthy adolescents (Cannistraro et al 2003). Finally, depressed adolescents were characterized by the failure of contingencies to influence latency and, by inference, cognitive control. This finding is consistent with the presence of motivation deficits, which neutralize the power of incentives to enhance behavior, such as attention engagement during errant prosaccades. Peak Velocity: Efficiency of Motor Execution Control Peak velocity indexes regulatory influences on already initiated actions, as opposed to latency, which indexes regulatory influences that occur before initiation of an action. Here again, contingencies improved the control of errors in healthy adolescents, with lower peak velocities during reward and punishment conditions relative to nonincentive condition (Figure 4). In contrast, contingencies showed no modulatory influence on peak velocity in anxious and depressed adolescents. This deficit in modulating the execution of an action under different incentive conditions might also reflect difficulty in mobilizing resources to correct an error. Whereas this control mechanism is tested here in the context of motor execution, it is conceivable that similar controls are exerted toward inhibiting unwanted thoughts or emotions. These results need to be considered in light of some limitations. First, the sample size was relatively small, particularly with respect to patients. Nonetheless, it was sufficiently large to detect differences among groups. Negative findings must be interpreted with particular caution. For example, a larger sample may provide sufficient power to detect hypothesized group differences on accuracy. In addition, the sample size did not permit examination of age and gender effects on www.sobp.org/journal

S. Jazbec et al the performance scores. These effects will be examined in future larger studies of healthy individuals as a first step. Second, we combined adolescents with MDD and comorbid anxiety disorders and adolescents with only MDD. Although the primary diagnosis in these adolescents was MDD, the inclusion of subjects with comorbid anxiety may have diminished potential differences between the anxiety and MDD groups. Of note, analyses not shown here, performed after excluding the three subjects with comorbid diagnoses, generated identical conclusions. Nevertheless, future studies of specificity might exclude subjects with comorbid MDD and anxiety. Third, three anxious patients also met criteria for mild ADHD. Although their performance did not differ significantly from that of the remaining patients with anxiety, it is not possible to evaluate the potential impact of such comorbidity on the performance of this task with only three subjects with comorbid ADHD. This is particularly salient given findings of impaired antisaccade performance in children with ADHD (Klein et al 2003). Fourth, although all anxiety patients had a history of GAD, they were heterogeneous with respect to other anxiety disorders. As a first study, we opted to include adolescents with more than one anxiety disorders because these conditions typically occur together (Pine et al 1998). The next step will be to examine these disorders separately. Fifth, it is possible that motivation to do well on the task was different in patients compared with healthy subjects because of their seeking treatment. It was made clear to these participants, however, that they did not have to complete this task to be in the treatment study. In addition, seeking treatment was initiated by the parents rather than by the adolescents. These circumstances mitigate the possibility of different sources of motivation to do the task between patients and healthy volunteers. Despite these limitations, we found that depressed and anxious adolescents demonstrated distinct patterns of incentiverelated modulation of the execution of unwanted reflexive movements. Anxious adolescents showed some modulation of response errors, particularly during the punishment condition, whereas depressed adolescents had difficulty modulating these responses during either reward or punishment conditions. These findings suggest different mechanisms underlying the incentiverelated modulation of actions as a function of diagnosis. The reward antisaccade task permits the parsing out of different processes that contribute to cognitive control in motor preparation and execution. Functional neuroimaging studies using this task will be able to refine our understanding of the mechanisms underlying these controls of action and their alterations in mood and anxiety disorders. Another important question is whether these deficits precede the manifestation of symptoms and thus constitute vulnerability factors, or contribute to symptom expressions, and respond to pharmacologic treatment. In the same vein, a developmental study of these deficits from childhood through adulthood could help to clarify the etiologic and functional significance of these alterations.

Austin MP, Mitchell P, Goodwin GM (2001): Cognitive deficits in depression: possible implications for functional neuropathology. Br J Psychiatry 178: 200 –206. Beuke CJ, Fischer R, McDowall J (2003): Anxiety and depression: Why and how to measure their separate effects. Clin Psychol Rev 23:831– 848. Brockmole JR, Carlson LA, Irwin DE (2002): Inhibition of attended processing during saccadic eye movements. Percept Psychophys 64:867– 881.

S. Jazbec et al Brown JW, Braver TS (2005): Learned predictions of error likelihood in the anterior cingulate cortex. Science 307:1118 –1121. Cannistraro PA, Rauch SL (2003): Neural circuitry of anxiety: Evidence from structural and functional neuroimaging studies. Psychopharmacol Bull 37:8 –25. Carpenter RHS (1988): Movements of the Eyes. London: Pion. Carter CS, Braver TS, Barch DM, Botvinick MM, Noll D, Cohen JD (1998): Anterior cingulate cortex, error detection, and the online monitoring of performance. Science 280:747–749. Costello EJ, Pine DS, Hammen C, March JS, Plotsky PM, Weissman MM, et al (2002): Development and natural history of mood disorders. Biol Psychiatry 52:529 –542. Dalgleish T, Taghavi R, Neshat-Doost H, Moradi A ??, Canterbury R, Yule W (2003): Patterns of processing bias for emotional information across clinical disorders: A comparison of attention, memory, and prospective cognition in children and adolescents with depression, generalized anxiety, and posttraumatic stress disorder. J Clin Child Adolesc Psychol 32: 10 –21. Davidson RJ, Pizzagalli D, Nitschke JB, Putnam K (2002): Depression: Perspectives from affective neuroscience. Annu Rev Psychol 53:545–574. Drevets WC (2001): Neuroimaging and neuropathological studies of depression: Implications for the cognitive– emotional features of mood disorders. Curr Opin Neurobiol 11:240 –249. Ehrenreich JT, Gross AM (2002): Biased attentional behavior in childhood anxiety. A review of theory and current empirical investigation. Clin Psychol Rev 22:991–1008. Everling S, Fischer B (1998): The antisaccade: A review of basic research and clinical studies. Neuropsychologia 36:885– 899. Fischer B, Weber H (1992): Characteristics of “anti” saccades in man. Exp Brain Res 89:415– 424. Godijn R, Theeuwes J (2003): Parallel allocation of attention prior to the execution of saccade sequences. J Exp Psychol Hum Percept Perform 29: 882– 896. Gotlib IH, Kasch KL, Traill S, Joormann J, Arnow BA, Johnson SL (2004): Coherence and specificity of information-processing biases in depression and social phobia. J Abnorm Psychol 113:386 –398. Holroyd CB, Coles MG (2002): The neural basis of human error processing: Reinforcement learning, dopamine, and the error-related negativity. Psychol Rev 109:679 –709. Holroyd CB, Nieuwenhuis S, Yeung N, Nystrom L, Mars RB, Coles MG, et al (2004): Dorsal anterior cingulate cortex shows fMRI response to internal and external error signals. Nat Neurosci 7:497– 498. Ikeda T, Hikosaka O (2003): Reward-dependent gain and bias of visual responses in primate superior colliculus. Neuron 39:693–700. Kaufman J, Birmaher B, Brent D, Rao U, Flynn C, Moreci P, et al (1997): Schedule for Affective Disorders and Schizophrenia for School-Age Children-Present and Lifetime Version (K-SADS-PL): Initial reliability and validity data. J Am Acad Child Adolesc Psychiatry 36:980 –988. Kent JM, Rauch SL (2003): Neurocircuitry of anxiety disorders. Curr Psychiatry Rep 5:266 –273. Klein CH, Raschke A, Brandenbusch A (2003): Development of pro- and antisaccades in children with attention-deficit hyperactivity disorder (ADHD) and healthy controls. Psychophysiology 40:17–28.

BIOL PSYCHIATRY 2005;58:632– 639 639 Kobayashi S, Lauwereyns J, Koizumi M, Sakagami M, Hikosaka O (2002): Influence of reward expectation on visuospatial processing in macaque lateral prefrontal cortex. J Neurophysiol 87:1488 –1498. Leigh RJ, Zee DS (1999): The saccadic system. In: Leight RJ, Zee DS, editors. The Neurology of Eye Movements. Oxford, UK: Oxford University Press, 90 –150. Luna B, Garver KE, Urban TA, Lazar NA, Sweeney JA (2004a): Maturation of cognitive processes from late childhood to adulthood. Child Dev 75: 1357–1372. Luna B, Sweeney JA (2004b): The emergence of collaborative brain function: FMRI studies of the development of response inhibition. Ann N Y Acad Sci 1021:296 –309. Munoz DP, Broughton JR, Goldring JE, Armstrong IT (1998): Age-related performance of human subjects on saccadic eye movement tasks. Exp Brain Res 121:391– 400. Munoz DP, Everling S (2004): Look away: The anti-saccade task and the voluntary control of eye movement. Nat Rev Neurosci 5:218 –228. Naranjo CA, Tremblay LK, Busto UE ?? (2001): The role of the brain reward system in depression. Prog Neuropsychopharmacol Biol Psychiatry 25: 781– 823. Olincy A, Ross RG, Young DA, Freedman R (1997): Age diminishes performance on an antisaccade eye movement task. Neurobiol Aging 18:483– 489. Pine DS, Cohen P, Gurley D, Brook J, Ma Y (1998): The risk for early-adulthood anxiety and depressive disorders in adolescents with anxiety and depressive disorders. Arch Gen Psychiatry 55:56 – 64. Pine DS, Klein RG, Coplan JD, Papp LA, Hoven CW, Martinez J, et al (2000): Differential carbon dioxide sensitivity in childhood anxiety disorders and nonill comparison group. Arch Gen Psychiatry 57:960 –967. Ridderinkhof KR, van den Wildenberg WP, Segalowitz SJ, Carter CS (2004): Neurocognitive mechanisms of cognitive control: The role of prefrontal cortex in action selection, response inhibition, performance monitoring, and reward-based learning. Brain Cogn 56:129 –140. Taghavi MR, Neshat-Doost HT, Moradi AR, Yule W, Dalgleish T (1999): Biases in visual attention in children and adolescents with clinical anxiety and mixed anxiety-depression. J Abnorm Child Psychol 27:215–223. Takikawa, Y, Kawagoe R, Itoh H, Nakahara H, Hikosaka O (2002): Modulation of saccadic eye movements by predicted reward outcome. Exp Brain Res 142:284 –291. Tucker DM, Luu P, Frishkoff G, Quiring J, Poulsen C (2003): Frontolimbic response to negative feedback in clinical depression. J Abnorm Psychol 112:667– 678. Wasserman EA, Miller RR (1997): What’s elementary about associative learning? Annu Rev Psychol 48:573– 607. Watanabe K, Lauwereyns J, Hikosaka O (2003a): Effects of motivational conflicts on visually elicited saccades in monkeys. Exp Brain Res 152:361–367. Watanabe K, Lauwereyns J, Hikosaka O (2003b): Neural correlates of rewarded and unrewarded eye movements in the primate caudate nucleus. J Neurosci 23:10052–10057. Wechsler D (1999): Wechsler Abbreviated Scale of Intelligence. San Antonio. TX: The Psychological Corporation. Williams JM, Mathews A, MacLeod C (1996): The emotional Stroop task and psychopathology. Psychol Bull 120:3–24.

www.sobp.org/journal