Design and Participants
We utilized a within-between-subject design. Ethical approval was received from the Ethics Committee of Shiraz University of Medical Sciences (IR.SUMS.REC.1400.815) and subjects signed written informed consent. Prior to data gathering, a power analysis was performed by G*Power software, suggesting a sample size of N=40 to achieve 95% test power to achieve an effect size of d=0.30 at an α error of 0.05 for mixed ANOVA. Seventy-seven subjects were included; however, 10 were excluded as they did not meet the inclusion criteria, and finally, 67 participants were regarded. The study included two groups: the Depressed Group (n=36) and the Control Group (n=31) .
The Depressed Group included those with Major Depressive Disorder (MDP) from psychological and psychiatric clinics in Shiraz, Iran. They met the diagnostic criteria for a current MDE within the last two weeks (following the DSM-5 criteria, assessed using the SCID-5-RV) and scored > 14 on the BDI-II (Beck et al., 1966). They aged between 18 and 50, had no color blindness (Ishihara, 1918), had at least a primary school education to ensure reading and comprehension of the instructions, and were not using antidepressants. The exclusion criteria encompassed having schizophrenia and/or other psychotic diseases, substance abuse, obsessive-compulsive disorder, bipolar disorder, and/or neurological conditions, like epilepsy brain injuries, and tumors. The depressed group had 36 people(age M=32.64 years, SD=6.07; 2 men; 34 women).
The Controls were 31 individuals and included the general public selected via social media platforms. All participants in this group had no DSM-5 disorder considering their answers on the SCID-5-RV. They had no psychiatric disorder or had not visited a psychiatric facility in the last six months (determined through a clinical interview and the GHQ-12) and were not using any drug to treat a psychiatric condition (mean age = 30.68 years, SD = 6.847; 9 men and 22 women) .
Measures
Structured Clinical Interview For DSM-V -Research Version (SCID-5-RV)
The DSM-5 diagnoses were assessed by this scale (First et al., 2015). It demonstrates strong psychometric features, such as in Iranians (Mohammadkhani et al., 2020). According to Mohammadkhani et al. (2020), the sensitivity index for MDP is 0.68, and the specificity is 0.75. In this study, a clinical psychologist administered the full SCID-5-RV to diagnose MDP and to ensure that the healthy control participants had no psychiatric disorders.
Beck Depression Inventory-II (BDI-II)
This scale with 21 items assessed the severity and existence of depressive symptoms in the past two weeks (Beck et al., 1996). The items are graded between 0 and 3 (total score: 0 - 63); scores above 14 indicate at least mild depression (Beck et al., 1996). The Persian BDI-II has established validity and reliability (Ghasemzadeh et al., 2005). In our research, Cronbach’s alpha was found to be 0.95.
General Health Questionnaire-12 (GHQ-12)
This scale with 12 items evaluated the general health status of subjects in the last four weeks. The items are graded between 0 and 3 (total score: 0 - 36), and higher scores indicate poorer health status (Goldberg, 1988). Yaghubi et al. reported that the GHQ-12 has strong psychometric properties in Iran. Here, Cronbach’s alpha was 0.93.
Cognitive Control Task
The Emotional Word Stroop Task (Williams et al., 1996) measured cognitive control exertion over emotional words (negative, positive, and neutral). Initially, a total of 120 words were selected for the stimuli pool and these words were piloted using 10 individuals with depression, who were asked to rank the words based on the valence (0 = completely unpleasant to 10 = completely pleasant). Finally, 22 negative words, 22 positive words, and 22 neutral words were selected for the Emotional Word Stroop task (see Appendix 1). Words were matched in length. Overall, each category of words was repeated three times across the main task. Participants were instructed to focus on the words’ ink color while disregarding the meaning of the emotionally charged words. We employed a previous paradigm for reward cues (Frömer et al., 2021) following ECV Theory (Shenhav et al., 2013). The subjects received rewards according to the reward cues presented, which included the reward amount (high and low = 150,000 and 20,000 Rials, respectively) and the efficacy level (high and low: rewards were completely based on subjects’ function and were not contingent on the subjects’ function but were randomly assigned, respectively) for each trial. The trials (Fig 1.) began with a fixation display lasting 1500 milliseconds, and then a reward cue was presented for another 1500 milliseconds. Next, the considered stimulus was shown for 1000 milliseconds. When the subject responded, a blank screen appeared for 800 milliseconds, and then feedback was displayed for 750 milliseconds, indicating the reward they would receive in the next trial. The interval between the two attempts was set at 800 milliseconds. To enhance the task’s difficulty, the response threshold for every trial was 750 milliseconds, although reaction times (RTs) were noted for up to 1000 milliseconds when the target stimulus appeared. The stimulus was provided at a screen center (12 inches) on a Microsoft Surface Pro 3 tablet, which was positioned 60 cm far from the participant.
Subjects started by completing 3 practice blocks. In the initial block, which included 16 trials, we displayed a square stimulus in yellow (255, 237, 0), blue (0, 5, 255), red (255, 0, 0), and green (0, 128, 0). Subjects familiarized themselves with the key-color mapping by putting pressure on the related keys on the keyboard (the F, D, K, and J keys were respectively assigned to green, blue, red, and yellow). The second practice block consisted of 20 trials, where subjects associated cues with different levels of efficacy and reward. Ultimately, subjects finished the third practice block, which comprised 32 trials and closely mirrored the actual task. Instructions for the incentive blocks are provided below:
In the next block, you again need to press the key associated with the color of the text on the screen. From now on, you will have the opportunity to get an additional bonus based on how you perform the task. You will be told on each trial how performance could affect your bonus. Before each word appears, you will see an image that tells you two things: (1) the amount of reward you could earn; and (2) whether or not your performance will determine if you get that reward. When you see one of the two images above, you can get a low (20000 Rials) or high reward (150000 rials) if you respond quickly and accurately. In other words, the blue hand shows that your reward is directly related to the speed and accuracy of your response. The gray bag represents a low reward and the pink bag represents a high reward. The two images above ALSO indicate that you can get a low or high reward, BUT the gray hands indicate that your reward will have NOTHING to do with how quickly or accurately you perform. Instead, these rewards will be determined randomly. As long as you provide some response on that trial, you have some possibility of getting a low (20000 Rials) or high (150000 Rials) reward. Although these rewards will be random, you will be just as likely to get a reward on these trials as the trials with the blue hands. (Frömer et al., 2021).
After each block of practice, the following statement was presented to ensure whether the practice was suitable for every subject: “Was the practice sufficient in this part? If this practice is enough, press the Y key. If you need more practice, press the N key”. Following completing the practice stage, Subjects were guided to perform the main task: “The main task is similar to the last practice, but it will not be practice. Every trial can affect your ultimate reward. At the end of the session, 10 trials will be chosen randomly and the total amount of money you earned across those 10 trials will be paid to you” (Frömer et al., 2021). The main task had a 198-trial block, which was counterbalanced between subjects. The JAVA language developed the experiment.
Procedure
The study took place in a single 90-minute session. Initially, subjects underwent a clinical interview using the SCID-5-RV, followed by the Ishihara task (Lezak et al., 2012) to evaluate color blindness. Those who met the eligibility criteria were allocated to either the healthy control group or the depressed group. Subjects filled out some assessments, such as the BDI-II, a demographic questionnaire, and the GHQ-12. Afterward, they engaged in the cognitive control approach. They then were provided with the rewards they earned during the task.
Statistical Analysis
Data analyses were performed by SPSS 27. The analyses assessed the average RT for correct responses (Accurate RTs) and the accuracy rate, calculated as the ratio of correct responses to total responses. A mixed repeated measures analysis of variance (ANOVA) with a 2 (Group: Depressed vs. Control) × 3 (Valence: positive, negative, and neutral) × 2 (Efficacy: low vs. high) × 2 (Reward: low vs. high) design examined the impact of expected efficacy and reward on cognitive control allocation. In this analysis, Group served as a between-subjects factor, while Valence, Efficacy, and Reward were treated as within-subjects factors. The dependent variables included average accurate RTs and the accuracy rate. Univariate outliers were identified by determining mean RTs that exceeded 3 standard deviations, leading to the exclusion of six participants (the RT data showed normal distribution when these outliers were removed) from the ANOVA, which used average RTs as the dependent variable. This left 27 subjects in the Depressed Group and 29 in the Controls for these assessments. Additionally, subjects (N = 22) who acted poorly (i.e., achieving < 60% accuracy on high efficacy trials) were excluded from the ANOVA that used accuracy rate as the dependent variable, resulting in 24 subjects in the Depressed Group and 21 in the Controls for these assessments. Effect sizes were reported as partial eta (medium: ηp2 = .06; small: ηp2 = .01; and large: ηp2 = .14) and Cohen’s d (medium: d = .09; small: d = .01; and large: d = .25) (Cohen, 1988).