Fifty right-handed participants (the age range from 20 to 55 years old) were recruited to the study. Patients (n = 26) were selected from Institute’s Outpatient Psychotherapy Clinic («A1 Clinic») and screened by one of three psychiatrists (YL, A.M., and L.A.) using the Structured Clinical Interview for DSM-V – (SCID, First et al., 2016) to identify cases of MDD. Those patients who were diagnosed with the MDD were included in the MDD group (American Psychiatric Association, 2013). The control (non-depressed) subjects (n = 24) were recruited following advertisement. All the participants were required to refrain from taking any psychotropic medication at the time of the study and were free from alcohol at least 3 weeks before. All the participants were interviewed with the Hamilton Rating Scale (HDRS-17, Hamilton, 1960), and completed the Beck Depression Inventory-II (BDI-II, Beck et al., 1996) and the Ruminative Responses Scale (RRS, Nolen-Hoeksema, 2000). All study participants provided written informed consent and the study was approved by the Ethical Committee of State Research Institute of Physiology and Basic Medicine (Novosibirsk, Russia).

Stabilometric recordings of CoPD were performed using the force-platform system linked to ST-150 software (sampling frequency of 33 Hz) (BioMera, Moscow, Russia).

Fine motor control is associated with tasks that typically involve some form of accurate manual manipulation. We administered actual-timed finger motor tasks (AMTs) in the form of sequential finger tapping – opposing each finger with the thumb in sequence, i.e., “index, middle, ring, little” comprises a set. Subjects reported the number of sets completed after the performance (Cambridge-Keeling, 2002). Before the test, the examiner gave verbal instruction while demonstrating the expected performance to ensure that the participant understood the instruction. Brief, untimed practice followed. After practicing the examiner said, “When I say ‘go’, do the same thing as fast as you can until I stop you.” The fluency of fine motor task performance was calculated by dividing the number of sets by the total amount of performance time as a number of sets per minute (set/min). In the mental motor task (MMT) condition, participants performed the same task, but instead of actually moving their fingers they were asked to imagine performing the movements. Participants were explicitly and repeatedly instructed to maintain the vividness of motor imagery throughout the experiment. All participants also calculated a number of sets to ensure that they were able to engage imagery and would be able to perform the task according to the instructions. Both actual and mental imaged motor tasks were performed in standing position.

The energy expended for CoPD was performed using three experimental tasks: (1) single-posture task as a quiet stance (ST), (2) actual dual-posture fine motor task (AMT); and (3) dual-posture mental/imagery version of the fine motor task (MMT). The EO and EC conditions were counterbalanced within the groups with the restriction that MMT always followed AMT. In the EO condition, the subjects looked at a fixed level target at a distance of 2 m. The participants were instructed to maintain an erect position for 30 s under each condition, with arms alongside the body and feet positioned on a schematic representing a 30° angle with respect to the anteroposterior direction. Two trials (each of 30 s duration) were performed for each condition. The results were averaged across trials. Those trials where artifact movement of eyes, head, arms or legs were video-recorded with webcam were excluded from the analysis.

For a trial of 30-s duration (sampling frequency of 33 Hz), the force-platform system was linked to ST-150 software (Biomera, Moscow, Russia), thus providing CoP series on the anteroposterior (Y) and mediolateral (X) axes assessment (Grohovsky and Kubryak, 2014). The dependent variables were calculated from the analysis of CoP trajectories. Trace length (L) in millimeters of CoP excursion and sway velocity (V in m/s) corresponding to the sum of the CoPD scalars divided by the sampling time were normalized to the participant’s height (Rocchi et al., 2004; Bazanova et al., 2017). The amount of kinetic energy (E) expanded for balance maintenance (mJ/kg) was calculated according to the formula

where m is the body weight, V the CoPD velocity, and n the number of discrete measurements conducted (Grohovsky and Kubryak, 2014).

To quantify subjects’ ability for executing two tasks concurrently, we calculated for each subject and task the absolute dual-task costs (McDowd, 1986), namely, the absolute dual-task energy costs index (i.e., ΔE), showing individual cost to the dual-tasking versus the single-posture task according to the formula: ΔE = (E_Dual-task - E_Single-task).

IBM SPSS statistic software was used to analyze data. Normality was ensured by Kolmogorov–Smirnov test. A three-way ANOVA with the factors of Group as between (Group 2: HC vs. MDD), and Vision (Vision 2: EC vs. EO) and Task (Task 3: ST vs. dual AMT vs. dual MMT) as repeated measures were performed and were followed by separate ANOVAs and planned comparisons. The Bonferroni correction was applied where appropriate.

In line with our primary purpose, correlation analyses were performed to assess interplays among clinical, posture, and motor fluency variables. A forward stepwise multiple regression analysis was used to determine if posture assessment variables under dual-tasking paradigms would predict clinical scores of MDD and compelling depressive rumination.

As shown in Table 1, participants in the depressed group reported significantly greater symptoms of depression and depressive rumination than those in the control group: HDRS-17, t(48) = 18.127, p < 0.001; BDI-II, t(47) = 13.92, p < 0.001; RRS, t(47) = 8.35, p < 0.001. Groups did not differ in age, t(48) = 0.099, p > 0.921, gender ratio, X² (1, n = 48) = 1.864, p > 0.397, education level, X² (1, n = 50) = 0.162, p > 0.685, or marital status, X² (2, n = 50) = 0.541, p = >0.760.

Correlational analyses (Pearson correlation) for the whole sample showed that fluency scores for the AMT and MMT tasks in both visual conditions negatively correlated with the HDRS-17, BDI-II, and RRS scores. Meanwhile, the E variables positively correlated with HDRS-17, BDI-II, and RRS score only in the single-task EC condition (r = -0.567, p < 0.004). On the other hand, the ΔE in AMT, ΔE in MMT both negatively associated with HDRS-17, BDI-II, and RRS scores (Table 3). The findings from regression analysis revealed that the ΔE in AMT and ΔE in MMT significantly predicted clinically relevant scores of depression and compelling depressive rumination (Table 4).

The rationale of manipulation with the visual input was that the EC condition makes the visual information unavailable, and therefore proprioceptive inputs must be predominantly used for an effective “automatic” posture control. When associated with the impaired integration of proprioceptive inputs in MDD, the essential “proprioceptive reweighting” process (Peterka and Loughlin, 2004) for maintaining posture control would be adversely influenced by divided attention performance (Barr et al., 2016; Deschamps et al., 2016). It is well documented that posture control abilities in humans depend on the capacities to detect the environment and visual, vestibular, and somatosensory inputs (see for review Newton, 1997; Forbes et al., 2015). More specifically, this is achieved primarily by a feedback-control mechanism that includes dynamic regulation of sensorimotor integration by the “reweighting of individual sensory channels” (Peterka and Loughlin, 2004). Vision and visual and spatial attention are among executive functions that are essential for successful posture stability and navigation through the environment (e.g., Albert et al., 1999; Nagamatsu et al., 2013). EC condition makes the visual information unreliable, and therefore remaining sensory inputs must be predominantly used for an effective “automatic” posture adjustment while a “load-compensation” mechanism attempts to modulate the magnitude of corrective torque needed to achieve good control of ST (Peterka and Loughlin, 2004; Bolbecker et al., 2011). As it was predicted, there was the significantly heavier impact of closed visual input in patients vs. HCs. Poorer posture stability in patients suggested an inability to compensate decreasing of sensorial inputs (vision), which could be mainly due to the PMR-related deficit of sensorimotor integration in depression (Bolbecker et al., 2001; Doumas et al., 2012; Deschamps et al., 2016). If so, it is associated with deficits in the integration of vestibular and proprioceptive inputs, the essential “reweighting of individual sensory channels” led to deficient posture control (Deschamps et al. 2015, 2016).

Before going to the dual-tasking and E costs for posture control, let us first have a closer look at the fluency data. Our main findings on these revealed a global slowness of both actual and mental movements in patients. Another distinctive feature was that fluency of imagined movements, as those of actual movement, was poorer in depressed patients, whereas controls acted equally well in both tasks. Our observation corresponds with previously described psychomotor deficits in unipolar depression (Bennabi et al., 2013, 2014; Liberg and Rahm, 2015; Thomas-Ollivier et al., 2017) and could be related to functional and structural changes within specialized brain circuits, associated with negative affect, cognitive and motor control, and action representation (Schrijvers et al., 2008; Walther et al., 2012; Liberg et al., 2013).

In line with our prediction, during execution of the actual or even imaginary concurrent motor task patients manifested reliable and statistically significant attenuation of energy expended for balance maintenance.

In an attempt to outline the possible mechanisms for such an improvement, we had accepted the following logic of reasoning.

On the one hand, according to the capacity model of attention (e.g., Kahneman, 1973), there is a finite reserve of cognitive resources, i.e., the more resources the given task requires, the fewer functional space that might be distributed to a secondary task is available. Dual-tasking cost could be a valid marker of frailty and a useful tool for identifying persons with signs of PMR (Lundin-Olsson et al., 1998). Simultaneous performance of an attention-demanding task (cognitive or psychomotor) during posture/gait tasks often leads to profound negative effects not only on gait and balance but also on the efficiency of cognitive task performance (Woollacott and Shumway-Cook, 2002; Al-Yahya et al., 2011). Such reductions in performance have been commonly interpreted as competition for attentional resources between the posture and cognitive task.

On the other hand, ruminative responding in MDD is defined as a recurrent, self-reflective, and unintentional focus on depressive symptomatology and its causes and consequences. A trait-like ruminative response style has been found to predict higher levels of depressive symptoms in depressed individuals, perhaps because of disrupted allocation of cognitive resources and increased recall and rehearsal of negative life events (Nolan et al., 1998; Kuehner and Weber, 1999; Nolen-Hoeksema, 2000). The depressive rumination overcrowds attentional resources, thereby accounting for the PMR signs of cognitive slowness and deficient psychomotor activity in MDD patients (Nolen-Hoeksema, 2000; Hamilton et al., 2011; Machino et al., 2014). Recent studies showed that of the two distinct and anti-correlating neuronal networks, i.e., the default-mode network (DMN) and the task-positive network (TPN), the first one may represent important neural substrates of depressive rumination (e.g., Hamilton et al., 2011). Performance of attention-demanding tasks activates the TPN circuits and reciprocally suppresses the DMN activity. During wakeful rest, the opposite pattern emerges, when the activated DMN lead to outbursts of depressive rumination in patients with MDD (Hamilton et al., 2011).

Let us put up together all of the above. Hypothetically, execution of the single-posture task in the EC condition increases the relative dominance of the DMN that propels depressive rumination accompanied by consumption of additional attentional resources. The dual-tasking “switches off” the DMN activity with inherent depressive rumination and redirect released attentional resource toward the secondary tasks and adaptive engagement of the TPN. As a result, rumination-linked posture deficit in patients with MDD is attenuated (Nolen-Hoeksema, 2000; Hamilton et al., 2011; Machino et al., 2014). Executing fine motor and ideomotor task performance might have an immediate positive effect on the limbic system, lead to changes in functionally connected remote areas, enhance activity of the hypo-activated regions and pathways, strengthen weakened frontostriatal coupling, and finally improve sensor-motor integration (Lohr et al., 2013; Baune et al., 2014; Roca et al., 2015).