Summary of journals

physical and mental health as they may hinder the degree to which complex decision-making tasks can be accurately and quickly executed on a daily basis, e.g., during driving, shopping, social interactions, etc. Moreover, due to the complex nature of depressive symptoms and related neural systems [4–5], evidence of cognitive control deficits in depression based on stan- dard cued response tasks is at best mixed [6], suggesting a need for more sensitive behavioral paradigms to assess cognitive control in depression.

Motivation plays an important role in cognitive control [7]. Motivation deficits in depressed individuals may lead to impaired performance in cognitive tasks. To understand the deficits of cognitive control in depressed individuals, it is important to examine the motivational pro- cesses that orienting the control process. Affective states not only affect cognitive but also moti- vational processes. For example, sad affect in depressed individuals may enhance mood- congruent information processing [8–9], which may result in reduced motivation to pursue rewards. Moreover, relative to healthy controls, depressed individuals pay less attention to pos- itive stimuli [8] and report lower positive affect in response to positive stimuli [10]. Finally, stu- dents with a range of self-reported depressed mood were slower to approach positive social cues [11]. Anhedonia (the decreased capability to experience pleasure and seek reward) is another essential symptom of Major Depressive Disorder (MDD; DSM-V; American Psychiat- ric Association, 2013) and is a core feature of reward-processing deficits in depression. It has been suggested that anhedonia might reduce the motivation to pursue future rewards and engage in pleasurable activities [12]. That is, a decreased ability to experience pleasure may reduce the incentive to act to seek pleasure or reward, leading to further decrease in positive affect and thus perpetuating reward-based motivational deficits in depression. Treadway & Zald 2011 [13] argued that depression may not only be affected by ‘consummatory anhedonia’, i.e. the deficits in the hedonic response to rewards, but also by ‘motivational anhedonia’, i.e. diminished motivation to pursue rewards. For instance, in a signal detection task under differ- ent monetary payoff conditions, clinically depressed individuals were shown to be less discrim- inative of and also less sensitive to monetary rewards, which lead to lower performance accuracy [14]. These reward-seeking deficits appear in the context of decreased recruitment of the nucleus accumbens and prefrontal cortex (PFC) upon exposure to rewards, which is believed to reflect impairment in attentional switching, regulatory processes [15–16] and in flexibility to alter reward-seeking behavior [17].

While motivational research often focuses on approach deficits in depression, a disturbance in avoidance processes may be equally important and relevant to cognitive control perfor- mance among depressed individuals [18]. For example, depressed patients learn faster to avoid risky gambles [19], and demonstrate faster motor response to withdraw from negative stimuli such as negative faces [20]. It has also been suggested that one of the characteristics of depres- sion is the increased difficulty to disengage from negative material [21–22]. Such a negative attention bias may lead to deploying greater cognitive effort to move away from undesired states (avoidance goals; [23]). This has been reported in a variety of studies, with evidence sug- gesting that depression is associated with more avoidant schemas and emotions [24]. Overall, these findings suggest that depression may result in an attenuated motivation to approach reward as well as greater sensitivity to and a higher motivation to avoid punishment. However, few studies [25] have attempted to quantify these motivational biases in depressed individuals, particularly in terms of their influence on motor-control in complex goal-directed tasks.

The Present Study The research reviewed above, as well as neural studies of frontal asymmetry, have highlighted the mapping of psychological motivational states to basic actions tendencies in healthy and

The Influence of Depression on Cognitive Control

PLOS ONE | DOI:10.1371/journal.pone.0143714 November 25, 2015 2 / 13

Alternatively, given evidence of decreased reward-seeking and approach motivation in depression [37], it could be argued that more severely depressed individuals have less motiva- tion to perform the task due to anhedonia [38]. That is, they may be less motivated to be precise about the stopping position, i.e. reduced effort to achieve high accuracy, and as a result stop further away. In addition, their increased stopping distance might be the result of attenuated motor control due to fatigue while still avoiding errors. However, our results show that the high depression group did not differ from healthy controls in terms of their positive control action values (first part of trial epoch), but rather applied higher deceleration when the car was getting closer to the stop sign. Thus, depressed individuals may be less efficient in incorporat- ing environmental cues that indicate increased need for cognitive control (e.g., seeing a stop sign in the distance), which provides evidence for their abnormal pro-active and reactive con- trol processes [1, 39], as found in the classic interference paradigms such as Go/NoGo and Stop-signal tasks.

In non-emotional Go/NoGo tasks, depressed individuals did not differ from healthy con- trols on Go trials [40], but exhibited an inhibitory control deficit on NoGo trials with increased rates of commission errors [41]. Similarly, without emotional cue or any performance feedback in our task, we showed that there was no significant difference in approaching (which resem- bles a ‘Go’ signal) between depressed and healthy individuals, whereas the difference appears in the avoidance of a stop- sign. It is important to point out that the stop-sign in our task is not the same as a ‘NoGo’ stimulus. Instead, it acts as a signal indicating the potential for perfor- mance errors. The increased stopping distance we observed in depressed individuals may be due to an increased aversive salience of this stimulus. While previous Go/NoGo paradigm focused primarily on discrete decisions with short reaction time frames, our approach provides rich data to examine how action is determined from the state of the environment (e.g. car dis- tance to stop-sign) at each time point. That is, examining the full motor trajectory from contin- uously action monitoring enables closer inspection of the interaction between Go-activation and NoGo-inhibition in continuous time. The recording of a continuous motor trajectory in effortful cognitive tasks has several advantages over traditional button press responses. First, the trajectory provides a rich behavioral response repertoire, which can be analyzed using sophisticated inverse optimization approaches. Second, approach and avoidance tendencies can be extracted across different time epochs relative to the target. Third, the distance to target provides an objective measure of performance, which can be related to or differentiated from the mood-driven subjective biases. Fourth, modification of the current paradigm can be used to modulate the level of cognitive control necessary for optimal performance. For example, additional stop-signal cues can be provided, which can be used to investigate how stimulus- response (S-R) contingency variations influences motor inhibition [42], and how stop-signal probability affect both proactive and reactive inhibition in a sop-signal anticipation task [43].

Recent neuroimaging studies support the notion that cognitive inefficiency for environmen- tal cues during cognitive control conditions may be due to increased emotional response to these cues, which may result in poorer initial action planning and delayed breaking action when closer to target. For instance, recent studies point to an association between depression severity and ACC activation to interference in standard Stroop paradigms ([44–46]). Such ACC hyperactivity may indicate greater conflict processing and thus noisier, less efficient motor control associated with greater action cost in the close to target range (i.e., delayed implementation of avoidant actions requiring greater late deceleration effort). Moreover, higher sensitivity to punishment and avoidant tendencies in depression has been linked to increased limbic reactivity (particularly the thalamus, amygdala and rostral ACC), thought to reflect stronger bottom-up processing of negative cues [47–49], and to decreased activity in the dorsolateral prefrontal cortex (DLPFC), which is likely to be reflect a reduced ability to

The Influence of Depression on Cognitive Control

PLOS ONE | DOI:10.1371/journal.pone.0143714 November 25, 2015 9 / 13

disengage from negative stimuli and related avoidant goals [17]. Thus, based on this literature, the present results may point to a stronger cognitive processing of and interference from visual cues that predict the need to implement cognitive control.

Another plausible explanation is that more depressed individuals are lacking fine and/or more precise motor control caused by impaired sensorimotor skills, which would lead to more inaccuracy and variability in stopping positions. That is, while there is somewhat mixed evi- dence of psychomotor retardation in depression [50–51], higher depression severity could result in slower sensory processing and slower motor initiation. We note, however, that we did not observe slowness in accelerating or decelerating actions in the depressed groups at any points during the trial epoch. It is possible that our findings may only generalize to the popula- tion studied here, i.e. college-aged students with a range of self-reported depression symptoms. However, this does not rule out that sensorimotor factors may play a role in more severely, clinically depressed individuals. Therefore, we plan to test the current paradigm in a clinical population with a larger sample size, and also incorporate individual differences in sensorimo- tor skills in the modeling of how goal setting and reward-processing influence motor control.

Interestingly, while a similar pattern of results was observed in the wall condition (i.e., stronger avoidance effects in more depressed individuals), the group difference was only observed for stopping distance and did not reach statistical significance for late deceleration. This was unexpected given that the wall condition may be perceived as involving a stronger punishment potential (from a sensory standpoint) if the target is missed, which should amplify rather than minimize avoidance. However, this finding needs to be interpreted with caution. First, on closer examination, this lack of significant group difference in the wall condition appears to stem from greater variability of maximum deceleration in the high depression group, who still had greater average peak deceleration than non-depressed subjects. Moreover the average peak deceleration in the high depression group appeared slightly shifted in time to even closer to target, consistent with less efficient motor control implementation, i.e., greater delay of avoidance actions (see Fig 4b). Second, the fixed task order (Stop-sign, Wall, Stop- sign, Wall) used here may have imparted a learning effect and reduced our ability to detect a group difference. Using a novel experimental paradigm to study the condition effect (Stop-sign vs. Wall), we initially designed the task such that the stop-sign and the wall have equal distance from the starting point. It was interesting to find out that the effect of the wall condition is not significant in the current task order. However, it is not that surprising, as subjects may have been over-trained in the stop-sign condition and thus fine-tuned their motor trajectory in the wall condition. It is important for future experiments to keep in mind such motor learning over time. In future studies it may be useful to vary the distances of the stop sign and the wall, and also counter-balance the task orders to eliminate the possible effect of adapted motor tra- jectory from prior experience. Finally, while our simulated driving task is a new paradigm and we could not directly estimate the effect sizes to be expected, a review of the literature with some form of goal-directed motor dependent measures indicated that likely effect sizes for depressed vs. control group differences ranged from medium to large [50, 52–54]. This suggests that, based on our group sample sizes we were likely underpowered for detecting small and medium effect sizes, which is a limitation of this study.

In conclusion, these results provide empirical evidence of altered motivational influences in motor control performance among individuals with depression. We found that those individu- als with the most severe level of depression stopped furthest away from the target and showed the greatest deceleration when approaching the instructed stopping target. Together, these results show that the influence of depression in modulating approach and avoidance motiva- tions and cognitive control may be more fluid and dynamic than expected, with the potential to influence different stages (action planning and execution) of a motor-control task. They

The Influence of Depression on Cognitive Control

PLOS ONE | DOI:10.1371/journal.pone.0143714 November 25, 2015 10 / 13