The stimulation from TMS is a stored charge that is delivered outside the scalp. The magnetic field generated by the TMS penetrates the skull, and electrical current is generated in the cortical neurons that are located in the area under where the TMS coil is placed [35]. This current then depolarizes the neuronal membranes by mimicking electrical stimulation, thus triggering action potential. Unlike neuromodulatory devices which have low temporal resolution despite similarity in spatial resolution as the TMS, TMS has been noted as a neurostimulatory device as a result of its relatively high temporal resolution [36]. The electrical current from TMS has shown to induce increase in action potentials that are sufficient enough to yield a major mechanic difference [35], suggesting a more long-term alteration in cognitive performance when used for the purpose of cognitive enhancement [32].

While the general mechanism of TMS has been established over the years as mentioned above, the optimal method and parameters to applying TMS for the purpose of cognitive enhancement has been long debated. During the earlier years of research in cognitive enhancement, TMS was more typically applied to stimulate the peripheral nerves [37]. However, over the last few decades, transcranial magnetic stimulation of the cerebral cortex has received much attention as it turns out the application of TMS over the motor cortex can directly stimulate the corticospinal axon [35]. Furthermore, the application of TMS over non-motor areas of the brain also became of interest as it extended possible research applications with respect to brain-behavior relationships [36]. With direct cortical stimulation, in contrast to peripheral nerve stimulation, both excitatory and inhibitory effects in cognitive function were observed depending on the frequency of the stimulation. In addition, as newer technologies emerged in the field of neuroimaging such as fMRI and neuronavigation, researchers are now able to observe the direct neural effects of TMS by working in tandem with such methodologies, potentially increasing the reliability of these studies [38].

Throughout the years, research also demonstrated that each type of TMS application elicits distinct physiological results. Among these, repetitive TMS (rTMS) has now become the most utilized methodology, due to its ability to alter cortical inhibition and excitation according to stimulation frequency [26]. In particular, recent research emphasize the application of rTMS for the purpose of cognitive enhancement. rTMS applies a certain number of stimulations per session to the area of stimulation over a prolonged period of time - typically between 20 to 40 minutes of treatment time - and its large number of condensed stimulation is what enables alterations in task performance. In the past, fast and repetitive exposure of rTMS was believed to have detrimental negative effects on cognitive processes, due to its disruptive nature [26]. However, studies later clarified that this is only in the case when rTMS is applied while simultaneously performing an online task. On the contrary, studies have indicated that rTMS that is delivered prior to a task or in short periods in between tasks can enhance the overall cognitive task performance [32], and therefore became popularized in the field of cognitive enhancement. Of the many studies that focus on the cognitive enhancing effects of TMS as discussed in the current review, the majority have used rTMS as the primary method of treatment. rTMS may therefore be a promising method of TMS treatment for various deficits and further improvements related to cognition, in both populations of cognitive deficits such as those with ADHD or dementia, as well as healthy populations.

One of the conventional ways to categorize rTMS is by frequency, which is typically labelled as low or high frequency. Low frequency rTMS (LF-rTMS) refers to a frequency of 1~4 Hz and is known to induce transient suppression or cortical excitability by depolarizing neurons. Studies that utilized LF-rTMS have shown that it indirectly affects emotion and behavior [38], and some improvement in language related tasks [39]. However, there are very few studies that have shown cognitive enhancement using LF-rTMS and as such, high frequency rTMS (HF-rTMS) is the most widely used technique for cognitive enhancement in clinical trials as well as healthy individuals.

HF-rTMS is defined as frequencies that are 5 Hz or higher, and has been found to enhance long-term cortical excitability which sustains up to a month subsequent to the treatment [40]. HF-rTMS was initially thought to have detrimental negative effects on cognitive processes due to its disruptive nature and high magnetic field intensity. However, studies later clarified that HF-rTMS that is applied prior to a task transiently enhances cortical excitability in a more effective manner as compared with LF-rTMS, and have further proven that it has more long-term effects in the cognitive domains than other types of TMS [25].

One of the factors to note with respect to TMS is its location-specific nature. The neurobehavioral effects from TMS can vary tremendously depending on the particular brain region that is stimulated. Therefore, noting the location of stimulation prior to treatment is essential depending on the treatment group, especially so in the clinical field. For instance, in the case of treatment for motor function in patients with Parkinson's disease, stimulation would more likely be applied to the primary motor cortex in order to treat for the deficits in motor function [41]. In contrast, in the case of patients with psychotic disorders including depression and OCD, stimulation would be applied within the vicinity of the prefrontal cortex to treat for psychological symptoms [42].

One of the key anatomical regions of interest in cognitive enhancement via TMS is the dorsolateral prefrontal cortex (DLPFC). The DLPFC is part of the Brodmann area 9 and is significantly involved in human cognition, particularly in various processing skills including working memory, attentional control, and processing episodic information [34,43]. Earlier fMRI studies have shown that the left DLPFC is activated during times of task performance, and its activation is increased in correlation with the difficulty of the task [34]. Since the discovery of the function of the DLPFC, many studies have undergone experimental studies to validate both the safety and efficacy as well as the cognitive enhancing effects of TMS to the left DLPFC. To name a few, numerous randomized, sham-controlled studies have found that applying rTMS to the left DLPFC enhances response inhibition [44,45,46,47], working memory [48,49], delayed improvement in logical memory [50] and verbal fluency [51]. Research have also diversified in methodology, and found that the simultaneous application of rTMS to both hemispheres of the DLPFC improved episodic memory, a finding which revealed that the underlying mechanism involves the alteration in frontal gamma oscillatory activity [52]. While there are several selective studies that found improvement in the cognitive domain by applying TMS elsewhere other than the DLPFC such as the left primary motor cortex to improve associate memory [53], most studies have found that stimulating the left DLPFC via rTMS yields the most effective cognitive enhancement.

In the case of rTMS and cognitive enhancement, studies have primarily focused on the domains of memory that are short-term in that they are not permanent changes to one's cognitive function (Table 1). Among the various types of memory, those that fit into this description in previous studies have been working memory [6,8,9,48,49,51,56,57,58,87], episodic memory [27,51,59,60], and logical memory [27,55].

Working memory refers to a cognitive system where information is temporarily stored for the purpose of processing, including complex cognitive tasks such as language comprehension, learning, and reasoning [55]. Throughout the years, studies have indicated that certain diseases show cognitive deficits in the working memory spectrum, and therefore affect one's ability to store and manipulate information online [55]. Within the field of schizophrenia, meta-analyses were performed to confirm this cognitive deficit in working memory in schizophrenic patients [56], and other analyses have also been performed in the field of depression [8], as well as ADHD [6,9]. While the definitive ways to assess working memory has not yet been stratified and there is a large variety of assessments that can be used, most trial studies have used similar tests to assess working memory. Within the working memory spectrum, many studies have used tasks including digit span, spatial span, and n-back test to measure verbal working memory, as well as spatial span and spatial working memory tasks from the Cambridge Neuropsychological Test Battery (CANTAB) for visual and spatial working memory. In the current review, 15 studies have been chosen collectively which have shown significant improvements in working memory using the above methods of assessment along with rTMS (Table 1). Among these, certain studies have suggested that the improvement in working memory in relation to HF-rTMS may yield greater effect when the working memory load is higher and be related to the reduction of excessive gamma oscillatory activity, as found in Barr and colleagues [57]. In addition, Levkovitz and colleagues (2011) have also noted that certain working memory domains may have greater effects depending on the area of the brain that is activated during the task performance, such as frontal lobe-related tasks in contrast to tasks that activate the striatal-parietal region of the brain [48]. While working memory can be further broken down into spatial working memory and verbal working memory, previous randomized controlled trials have shown significant improvements in both domains of working memory after the treatment of rTMS [58].

Although limited in number, several studies have also indicated improvements in the three pinnacle memory domains, which are episodic, semantic and autobiographical memory after TMS treatment. While these memory domains are primarily related to the medial temporal lobe and the hippocampus as suggested in previous studies [59], rTMS treatment to the DLPFC have yielded significant results. Episodic memory, in particular, have shown to be reduced in individuals with cognitive deficits such as those with amnestic mild cognitive impairment [60], and have been of interest to best prolong and retain in the general population. Gagnon and colleagues (2011) have shown that healthy individuals with normal cognitive abilities have shown increased episodic memory as measured via Remember/Know recognition task, although this was done so using a paired-pulse TMS methodology [23]. Furthermore, Fitzgerald and colleagues (2003) have shown improvements in semantic memory as well as autobiographical memory after the treatment of rTMS, where both low and high frequency rTMS applied to the left DLPFC for 4 consecutive weeks have yielded significant improvements [51].

Another beneficial factor that has been suggested as a result of rTMS stimulation is the enhancement of cognitive processing. Cognitive processing is a domain of cognitive function that is distinct from memory, as it is less related to knowledge and more towards non-executive functioning such as selective attention. While the idea of enhancing one's cognitive abilities is generally familiarized as enhancement regarding intellect such as memory domains, processing skills are also a key factor that is significantly associated with intellect. Studies have shown that individuals with higher intelligence quotients tend to have faster processing speed [61], which may arguably be affected by the neural system. Therefore, the electrical current that stimulates the cortical neurons via rTMS may affect the neural pathways in such a way that it leads to a more efficient neural network in terms of processing information.

Within recent years, several studies have demonstrated the effects of rTMS in the improvement of cognitive processing (Table 2). In particular, many of these studies have used the Stroop interference task as the method of assessment for cognitive processing, as it has long been noted as a reliable way of measuring top-down attentional control including attentional vigilance and inhibition [31,44,45,46,47]. While these studies were based on many different target groups including healthy individuals, those with depression, and those with cognitive impairment, all the studies have shown consistency in their findings in that HF-rTMS to the left DLPFC increased the overall performance in the Stroop interference task. The collective findings suggest that the effects of rTMS do not necessarily differ between individuals based on intellect or illness, but rather largely depend on the application process of TMS including the area of stimulation and frequency.

In addition, there are many studies that support the effects of rTMS in enhanced cognitive processing using lesser conventional methods of assessment for cognitive processing including the assessment of psychomotor speed using the Motor Agitation and Retardation Scale [62], the Divided Attention Task [38], and the Letter Cancellation Test [47], all of which demonstrated a significant improvement in task performance.

Previous studies on cognitive enhancement using rTMS can also be generalized into categories of sample groups. While the application of rTMS has been FDA-approved for the target population of individuals with depression, studies have also shown that healthy individuals show improved cognition after rTMS. In particular, Vanderhasselt and colleagues (2006; 2007) have shown consistent results of the effects of rTMS on the DLPFC in improved performance on the Stroop interference task. The two studies support the efficacy of rTMS in higher performance of attentional control for individuals without any particular neural deficit, and further suggest that the effects are significant regardless of the hemisphere of stimulation [45,46]. Additional studies demonstrated improved attention as a result of rTMS in healthy individuals, using other measures for attentional control including Conner's Continuous Performance Task [31].

Furthermore, numerous rTMS studies on healthy individuals also demonstrated various domains of memory enhancement, including recognition memory and working memory [27,49]. While rTMS on healthy individuals has not yet proven to improve long-term memory or autobiographical memory as found in a depression model [51], improved recognition memory within healthy individuals may support previous theory that recognition is largely affected by changes in the hippocampus of the brain [63,64]. Given that the prefrontal cortex is stimulated in rTMS treatment, coupled with the fact that the prefrontal cortex interplays with the hippocampus in human memory [65], rTMS treatment to the DLPFC may indirectly affect the hippocampus in such a way that enhances recognition memory of healthy individuals. Future studies that investigate the connectivity and activation of brain networks that include the area of stimulation in rTMS may contribute to the delineation of the neural pathways of rTMS treatment.

In contrast to healthy individuals with a normal range of cognitive function, there is a much higher number of studies that demonstrate the efficacy of rTMS in individuals with disorders including depression and PTSD (Table 1 and 2). While many studies have reported cognitive deficit as a secondary symptom of depression and/or related emotional distress including PTSD, the precise mechanism that underlie this effect is yet to be elucidated. One of the theories behind this effect that is most agreed upon is the underactivation of the prefrontal cortex of the brain as a response to cope for the psychological distress [66]. The underactivation is coupled with lower cerebral blood flow to the prefrontal cortex, which results in the disturbance of its key function which is higher thinking and executive functioning. The application of rTMS to the left DLPFC may act as an enhancer to the deficit of the prefrontal areas including activation and cerebral blood flow, and has shown to improve memories of those under depression and PTSD in various domains including episodic memory, spatial memory, verbal fluency, and executive function [51,67,68,69,70]. rTMS treatment in depression has also demonstrated improved attentional control including the trail-making test and Stroop interference task [7,45,46,67,70,71,72,73,74]. While cognitive enhancement through rTMS in models of depression and PTSD has been consistent, further neuroimaging studies that investigate the physiological changes of the brain in response to rTMS may benefit in the understanding of the precise neural mechanisms underlying the magnetic stimulation of the DLPFC.

The current review demonstrates that rTMS treatment has a significant effect in cognitive function across various groups including healthy and diseased populations. However, it is noteworthy that the precise neurological pathway that underlies the effects of rTMS has yet to be elucidated. While the associations between brain regions and their functional roles have been investigated throughout the history of neuroscience, research that directly investigate the neurological pathways in the perspective of rTMS treatment is still limited. Further studies that directly observe the alterations in certain brain network connectivity related to the regions of stimulation in rTMS may help understand its mechanisms, along with the development of novel rTMS parameters. Previous studies in cognitive neuroscience have demonstrated that certain brain networks and their inter-connectivity with other regions of the brain strongly affect cognitive outcomes including memory. A recent study by Bridge and colleagues (2017) have shown the significant association between the hippocampus and the frontoparietal network (FPN) of the brain on retrieval memory [75]. This inter-connectivity relationship between brain regions and networks may help explain the neural pathways that underlie rTMS stimulation, as the DLPFC is one of the main components of the FPN. The magnetic stimulation to the DLPFC may strengthen the connectivity within the FPN, and consequently increase the inter-connectivity strength between the FPN and hippocampus, yielding an enhancement in memory. This particular mechanism may also be generalizable to other memory domains, and further studies that investigate this process according to different subtypes of memory may help elucidate the validity and generalizability of this potential pathway.

Furthermore, studies have also noted the significant role of the precuneus in cognitive function. The precuneus is another major component of the FPN alongside with the DLPFC, as well as a key modulator of the default mode network (DMN) of the brain [76]. By contributing to two of the major brain networks, the precuneus is involved with a number of important functions including performance of integrated tasks, eye movement, imagery, and self-processing operations [76]. Studies including Utevsky and colleagues (2014) have demonstrated the significance of the precuneus in modulating the default-mode network (DMN), particularly by the optimization or enhancement of the DMN [77]. This optimization of the DMN activity was also found to be in correlation with enhanced resting-state cognition, which can also be interchangeable to attentiveness [77]. Considering these findings, rTMS treatment to the DLPFC may stimulate and optimize the activity of the precuneus within the FPN, resulting in increased connectivity strength between the precuneus and the DMN. The heightened attentiveness that results from this potential neural pathway may explain for the increased processing speed as summarized in the current review (Table 2). A detailed schematic diagram of the two potential neural pathways that underlie as part of the mechanisms of rTMS is also provided (Fig. 1).

Despite these potential mechanisms of TMS, however, it is noteworthy that contradictory findings have also been present. A portion of the previous research that used similar if not identical parameters to those that demonstrated significant cognitive enhancement via rTMS treatment, also demonstrated insignificant findings (Table 3). As many of these studies used both parameters as well as outcome variables that are similar to those described in Tables 1 and 2, this suggests that the cognitive enhancing effects of rTMS should not yet be considered as valid.