Although Parkinson’s disease (PD) is mainly considered to be a motor disorder, the importance of non-motor symptoms, such as depression, psychosis and cognitive deficits, has been increasingly recognized (Watson et al., 2010; Pellicano et al., 2017; Schapira et al., 2017). Most of these features have been shown to critically influence patients’ quality of life even in the early stages of the disease (Schrag et al., 2000; Antonini et al., 2012; Pellicano et al., 2015). Cognitive deficits, extensively documented and often defined as frontal type executive dysfunctions (Gotham et al., 1988), include impairments of verbal fluency (Obeso et al., 2012), deficits in working memory (Lee et al., 2010), and attention deficits both in the early and moderate stages of the disease (Zhou et al., 2012). These deficits remain difficult to manage with current pharmacological medications, which are mainly directed at addressing motor dysfunctions. While levodopa treatment might only partially restore some cognitive functionalities (Lange et al., 1992; Burn, 2015), a number of these are not affected by dopamine-related treatments and other neuromodulators such as acetylcholine, noradrenaline, and serotonin, might be involved (Biundo et al., 2016; Caligiore et al., 2016; Yang et al., 2016).

Cognitive dysfunctions in PD have also been treated through non-pharmacologic therapies. These approaches are mainly based on cognitive rehabilitation and physical therapy (Murray et al., 2014; Dancis and Cotter, 2015). Cognitive rehabilitation includes four domains of practice (Sinforiani et al., 2004; Biundo et al., 2017): (i) educate the patient about their cognitive weaknesses and strengths; (ii) help to develop lost cognitive skills through retraining; (iii) develop compensatory strategies; and (iv) use the three skills developed through (i–iii) to enhance function in life activities. In PD patients this technique has shown significant, albeit modest, improvement in cognitive domains (París et al., 2011; Cerasa et al., 2014). Physical exercise, which is typically used to address some motor issues of PD (e.g., bradykinesia, postural balance problems) (Tillerson et al., 2003; Ridgel et al., 2009; Allen et al., 2011) has also been identified as a possible treatment for cognitive deficits (Petzinger et al., 2013; Pellicano et al., 2015). Clinical studies have shown that various types of exercise, including aerobic, resistance and dance can improve cognitive functions, learning and memory in PD patients, although the optimal type, amount, mechanisms, and duration of exercise are unclear (Murray et al., 2014). It has also been shown that a combined action of physical therapy and transcranial direct current stimulation may lead to an improvement in cognitive functions, that is, frontal abilities and/or global cognitive ability scales (Manenti et al., 2016). Motor imagery and virtual reality are also two therapeutic approaches that make use of cognitive function to enhance movement and cognitive aspects of people with PD (Mirelman et al., 2013a). For example, it has been shown that an intervention that combines treadmill training augmented by virtual reality reduces fall risk, improves mobility and enhances cognitive function in a diverse group of older adults (Mirelman et al., 2013b).

Recently, the efficacy of two new non-pharmacological PD rehabilitation approaches has been explored: action observation therapy (AOT) and dual-task rehabilitative training. AOT is based on the evidence showing that during observation of a movement, the related action representation “resonates” (i.e., it is reactivated) in our motor system (Rizzolatti et al., 2001). This motor resonance can drive the process of understanding the intention (goal) of the agent performing the action through a facilitatory effect on the motor pathways (Buccino et al., 2001; Wheaton et al., 2004). These processes can drive the learning and acquisition of motor skills in analogous ways as physical exercise (Porro et al., 2007; van der Helden et al., 2010; Higuchi et al., 2012). During AOT, participants are typically required to carefully observe videos showing actions that they later have to execute. It has been shown that AOT can lead to improvements in the performance of movements in PD patients involved in single session experiments (Abbruzzese et al., 2015; Caligiore et al., 2017). In addition, two studies have shown that AOT-based long-term rehabilitation programs, involving repeated sessions spanning over weeks/months, could provide some benefits for PD patients motor recovery (Pelosin et al., 2010; Buccino et al., 2011). To date, AOT has never been used to deal with cognitive deficits in PD.

Dual task requires participants to perform two tasks simultaneously, which interfere with each other, such as engaging in a cognitive task while executing a motor task (Pashler, 1994; Woollacott and Shumway-Cook, 2002). PD patients commonly have difficulty in performing dual task (Benecke et al., 1986; Castiello, 1997). These difficulties can be observed in the performance of two motor tasks, or two cognitive tasks, or combined cognitive and motor tasks (Brown and Marsden, 1991; Woollacott and Shumway-Cook, 2002; Kelly et al., 2012; Nocera et al., 2013). The neural causes of this problem have been related to limited attentional resources, defective central executive function, and less automaticity (Wu and Hallett, 2008; Wu et al., 2014). Dual-task rehabilitative training in PD patients uses the interference between cognitive and motor tasks to obtain functional improvements. Typically, the concurrent performance of motor and cognitive tasks in dual task positively affects the performance of one of them. For example, it may improve gait velocity, stride length, balance or some aspects of cognition (usually mental flexibility and processing speed) (Woollacott and Shumway-Cook, 2002; Silsupadol et al., 2006; Fritz et al., 2015).

In this article, we propose that AOT, traditionally used to improve motor skills in Parkinsonian patients, may also be used to improve their cognitive abilities if it is used within a dual task framework. We articulate our hypothesis by pivoting on recent literature and on preliminary results we obtained through a pilot study we designed to test the efficacy of a long-term rehabilitation program for therapy that, for the first time, uses AOT within a dual task framework to address cognitive impairments in PD. The results of the experiment show that using AOT within a dual task framework leads to relevant improvements in cognitive aspects related to working memory (verbal and visuospatial memory) and attention. In contrast, no significant improvement in motor behavior was found, which is a typical result when using AOT in isolation (Buccino, 2014; Abbruzzese et al., 2015; Caligiore et al., 2017). We propose that AOT with dual task may help patients to deal with the difficulty of filtering out irrelevant information and with the tendency that they have of losing the task goal, which are both features that characterize cognitive deficits in PD (Lee et al., 2010). In particular, we hypothesized that using AOT with a dual task may engage and train goal-centered executive functions (Diamond, 2013) through two interacting mechanisms. First, AOT stimulates goal activation based on motor resonance mechanisms activated within the mirror neuron system (Rizzolatti and Craighero, 2004; Craighero et al., 2007; Thill et al., 2013). Second, during AOT the patient also performs a mathematical or lexical task. The accomplishment of this second goal tends to engage working memory and attention systems that are also involved in the motor tasks (Silsupadol et al., 2006). If the tasks are sufficiently motivating, the goals might be persistently maintained active and guide the performance of suitable actions to accomplish them, notwithstanding the possible cross interference and distraction between them. This exercise results in a cognitive effort that leads to the strengthening of the executive functions.

The executive processes related to goal activation and maintenance are particularly important for PD cognitive dysfunctions and their treatment because they strongly involve the regulation of the brain mechanisms underlying goal-directed behavior by the dopamine and basal-ganglia systems (Balleine and O’Doherty, 2010; Mannella et al., 2016). Indeed, both empirical evidence and computational models show the importance of dopamine and nucleus accumbens for the activation of goals and their persistent permanence in time (Redgrave et al., 2010; Fiore et al., 2014; Floresco, 2015). Important for therapy effectiveness, these processes can also be enhanced by the release of dopamine caused by the engaging and novel features of the tasks used for training (Lisman and Grace, 2005; Mannella et al., 2013).

The progression of the PD tends to cause, on average, a progressive deterioration of motor and cognitive capabilities in patients (Antonini et al., 2012; Caligiore et al., 2016; Schapira et al., 2017). The benefits of therapeutic interventions should hence show an interruption of the deterioration trend and, when possible, an improvement of those capabilities. Such benefits might manifest right after the intervention, and hopefully persist in a later follow-up monitoring. In some other cases the beneficial effects might be detected only in the follow-up tests if they require brain consolidation mechanisms to manifest in behavior (Pelosin et al., 2010).

The results show that after the intervention there are significant improvements in both short-term and long-term verbal memory (Rey Auditory Verbal Learning Test), in long-term visuospatial memory (Rey-Osterrieth Complex Figure Test), and in some attentional/focussing aspects (Stroop Tests). Interestingly, for the Rey Auditory Verbal Learning Test, for both short-term memory test trials, and for the long-term memory test trials there are no significant differences between post-test and follow-up conditions. This indicates a persistence of the effect of the treatment 1 month after it ends.

Overall these data suggest that using AOT with a dual task can be effective in the treatment of PD frontal deficits, in particular, to develop improvement in working memory and attention. In particular, the results support our hypothesis for which the use of AOT with a dual task may foster goal setting and maintenance in the presence of distractors by training the working memory and attention executive systems. The results might rely on two mechanisms. First, AOT supports goal focusing through the mirror neuron system whose functioning involves both cortical (Rizzolatti et al., 1996; Fogassi et al., 2005) and sub-cortical areas (Caligiore et al., 2013; Bonini, 2016; Bruni et al., 2018). AOT is based on a motor resonance mechanism, reproduced by the mirror neurons firing (di Pellegrino et al., 1992; Rizzolatti et al., 1996; Buccino et al., 2001; Mukamel et al., 2010) when the participant observes another agent performing a goal-directed action, such as, for example, grasping an object (Rizzolatti and Craighero, 2004). Johnson-Frey et al. (2003) have shown through an fMRI study that the human frontal mirror regions are preferentially activated by the sight of images showing a hand grasping an object compared to a hand touching it. This indicates that mirror neurons tend to encode action goals such as the terminal state resulting from a grasping action (e.g., a certain relationship between the hand and the object). Along the same lines, Fogassi et al. (2005) found that some, but not all, mirror neurons in the parietal cortex of monkeys are selective to ultimate (high-level) goals that a given action contributes to obtain (e.g., “grasp to eat” vs. “grasp to place”). Overall, these data indicate that the mirror system involves the representation of goals at different levels of abstraction (Rizzolatti and Craighero, 2004; Craighero et al., 2007; Thill et al., 2013). Thus AOT, leveraging on the motor resonance mechanism reproduced by the mirror neurons firing, might be a relevant means usable in treatments to activate the participants’ goals.

Second, the additional exercises performed during AOT through the dual-task procedure (e.g., math operations, listing alphabetic letters), challenge the working memory functions. In particular, they train the participants’ capacity for maintaining the activation of two goals, one related to the cognitive task and the other related to the motor task. In this way, the role of cognition and concentration is fostered to the detriment of the performance of motor tasks (Woollacott and Shumway-Cook, 2002; Silsupadol et al., 2006).

This employment of AOT within a dual-task setting might be particularly valuable and effective for the treatment of PD symptoms as the activation and focussed maintenance on specific goals strongly relies on the effective functioning of the basal ganglia and dopaminergic systems (Redgrave et al., 2010; Fiore et al., 2014; Floresco, 2015). Given this involvement of the dopaminergic system, the treatment is also expected to be more effective for therapy if the training involves actions that are engaging and have a high functional value for the participants as this results in a stronger stimulation of the dopaminergic system and ventral basal ganglia (Redgrave and Gurney, 2006; Dagher and Robbins, 2009; Baldassarre et al., 2013). In this respect, most participants of this study informally reported that the training was at the same time challenging but also doable. This balanced level of challenge led them to a high engagement that might have played a relevant role in the positive outcome of the treatment (future work should further investigate these motivational aspects).

These mechanisms may also explain why AOT together with the dual task were able to facilitate cognitive improvement but did not lead to a significant motor improvement, as usually found when AOT is used alone. Indeed, the approach used might have in particular focussed the participants’ training on goal-related cognitive processes rather than on motor ones. This perspective is coherent with influential proposals highlighting how PD patients may be facilitated to operate in a goal-directed control mode (Redgrave et al., 2010) and also with literature suggesting that goal-based exercises may be effective in addressing deficits in PD (Petzinger et al., 2013).

While the results achieved support the idea that using AOT together with a dual task could provide a new way to treat cognitive deficits in Parkinson, more work is needed to design studies that compare the effect of combined AOT with dual tasks in control groups (e.g., performing only one activity, namely AOT, or the dual task, or motor exercise). This, together with the low sample size of PD patients, are the main reasons for which we propose this article as a pilot study. These further investigations could allow to study the role of each single component of the therapy (AOT/dual task/motor exercise). For example, it might be possible that the dual task training in isolation could be enough to improve cognitive measures. Moreover, in light of the absence of outcomes in the motor domains obtained in the current experiment, it might also be possible that the dual task component negated the potential benefit of the AOT component if the performance of such cognitive task was prioritized over the motor activity by the participants [see Kelly et al. (2012), for discussion of dual task prioritization in PD]. Overall, the experiment illustrated here shows that the proposed treatment had clinically relevant effects on cognition, but further experiments are needed to understand the contribution of its different components, or of their interaction, to the beneficial effects obtained.

Several single session experiments (Abbruzzese et al., 2015; Caligiore et al., 2017) and two studies based on a long-term rehabilitation programs (Pelosin et al., 2010; Buccino et al., 2011) have demonstrated the benefits of AOT in PD motor rehabilitation. This research demonstrates for the first time, through a long-term rehabilitative intervention, that AOT could also lead to the development of cognitive improvement in PD patients if used within a dual task framework. We suggest that this happens because AOT with a dual task trains PD patients to better deal with the difficulty of filtering out irrelevant information as well as with the tendency of losing focus on pursued goals, which are both features that characterize the cognitive deficits of PD (Lee et al., 2010) and strongly depend on the dopamine system (Fiore et al., 2014; Floresco, 2015).

DC, MM, FEP, GS, and GB conceived the presented idea and developed the theory. DC, MM, FEP, GS, FA, and GB designed the experiments. MM, LR, AF, and FP performed the experiments. DC, MM, LR, and GB analyzed the data. All authors verified the analytical methods, discussed the results, and gave feedback on the manuscript. DC wrote the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.