Aging is accompanied by a progressive decline in brain structure, physiology, and function, with increased risks of developing various neurological and cognitive disorders, e.g., dementia and Alzheimer’s disease (Blinkouskaya et al., 2021). Cognitive functioning has relevant practical and direct impacts on older individuals since it plays a crucial role in their quality of life. Therefore, studies into whether there are strategies to counteract cognitive decline in aging are of utmost importance.

Among cognitive domains, episodic memory (EM) exhibits the largest degree of age-related impairment, with an augmented decline usually observed after the age of 65 (Rönnlund et al., 2005; Guillaume et al., 2009; Vestergren and Nilsson, 2011). EM is a part of the explicit long-term memory that stores information about personal life, such as temporally dated episodes or events (Tulving, 1984). Therefore, it enables a person to remember directly experienced events, making it possible to be consciously aware of a well-contextualized past experience (Tulving, 1993).

Anodal transcranial direct current stimulation (atDCS) has attracted substantial attention in recent years as a tool to improve memory performance, as well as other cognitive functions (see Goldthorpe et al., 2020; Sandrini et al., 2020; Huo et al., 2021; Indahlastari et al., 2021; Sanches et al., 2021; Siegert et al., 2021, for recent reviews). In fact, there is evidence in young participants demonstrating that its application can modulate ERPs elicited in an EM task (Lu et al., 2015). tDCS consists of the application of a weak electrical current (typically 1–3 mA) through two or more electrodes placed over the scalp that reach the cortex and modulate neuronal transmembrane potentials (Purpura and McMurtry, 1965; Nitsche and Paulus, 2000). Despite the simplistic view that atDCS increases neuronal excitability and enhances behavioral performance (whereas cathodal tDCS decreases neuronal excitability and worsens behavior), the scientific literature clearly demonstrates that tDCS does not always lead to such linear neuropsychological and behavioral outcomes (Fertonani and Miniussi, 2017).

The results regarding the cognitive effects of tDCS in aging are mixed, as the administration of atDCS has not consistently resulted in the enhancement of cognitive functions in older individuals. Some tDCS studies have reported memory improvement in healthy older adults when stimulating the prefrontal cortex (Manenti et al., 2013, 2017; Sandrini et al., 2014, 2016, 2019; Medvedeva et al., 2019; Huo et al., 2020) or the left temporoparietal cortex (Antonenko et al., 2019). On the other hand, other studies applying tDCS over the prefrontal cortex reported no effects or even a worsening in aging (Leach et al., 2016, 2019; Peter et al., 2019; Habich et al., 2020a,b). Therefore, it is clear that brain responsiveness to tDCS shows high inter-individual variability.

Several studies have identified age as a key modulating factor of neural and behavioral effects of tDCS (e.g., Antonenko et al., 2016, 2018; Fiori et al., 2017; Martin et al., 2017). Considering the evidence for age-related structural and functional reorganization, tDCS may operate, at different ages, upon different neural processes (Antonenko et al., 2019). Importantly, these studies suggest that results obtained in an age group may not be transferable to another age group. However, the current literature does not allow us to understand the role of age as a moderating factor of the interindividual variability of tDCS effects within the aged population itself. First, the available literature considered the older population in a wide range of ages covering over 30 years with little or no investigation of the large inter-individual variability within this range [e.g., in Antonenko et al. (2019) authors included participants from 50 to 80 years; in Fiori et al. (2017) and Leach et al. (2019) 60–80 years; in Leach et al. (2016) 60–90 years]. Second, the studies on age-mediated effects of tDCS have focused exclusively on the comparison of older (usually in a range from 50 years and older) with younger (aged 18–35) adults. Brain responsiveness to tDCS among older individuals may even vary more than among younger individuals due to considerable age-related inter-individual differences in cognitive decline and brain structure alterations in this population (Craik et al., 1994; Mungas et al., 2010; Reuter-Lorenz and Park, 2014; Antonenko et al., 2019). Additionally, susceptibility to plasticity-inducing mechanisms is affected in the older adults (Brosnan et al., 2018; Antonenko et al., 2019), and consequently, the brain response to tDCS may be altered. Hence, the heterogeneity of findings on the effects of tDCS on memory in aging further highlights the complexity of the underlying mechanisms (Antonenko et al., 2019) and requires monitoring for age-related inter-individual differences.

In this scenario, middle-aged adults are of major interest, as it may be the stage when the first changes in brain activity occur even in the absence of behavioral decline (Guillaume et al., 2009). To the best of our knowledge, only one study compared, in a posteriori analysis, middle-aged (<63 years) and older (>63 years) adults, pointing to a larger reduction in GABA concentrations after tDCS over the sensorimotor region in middle-aged adults. However, no behavioral data that can distinguish group performance were collected (Antonenko et al., 2017).

In sum, the understanding of tDCS effects on the aged brain is still inadequate (Antonenko et al., 2019), although older individuals may be among the primary targets of interventions. Therefore, more studies comparing middle-aged with older adults would help to clarify how tDCS effects on EM may change depending on the subject age.

The mechanisms involved in EM have been extensively studied using the event-related potentials (ERPs) technique, identifying three main components: (1) the Early Frontal effect (EF), a positive component that usually appears at ~300–500 ms and is maximal at frontal locations, thought to reflect familiarity-based recognition processes (Curran, 2000; Rugg and Curran, 2007); (2) the Parietal Effect (P), a subsequent positive component that appears in the ~400–800 ms interval at left parietal regions and is thought to index recollection-based recognition processes (Rugg and Curran, 2007); and (3) the Late Frontal effect (LF), a later positive ERP component that has been associated with post-retrieval monitoring and evaluation processes, and is likely linked to executive function of the prefrontal cortex (Friedman and Johnson, 2000; Rugg, 2004; Hayama et al., 2008). To our knowledge, no studies have evaluated how tDCS modulates ERP components associated with EM in a healthy aged population.

In the present study, we aimed to delineate the trajectories of atDCS effects applied over the left dorsolateral prefrontal cortex (DLPFC) during the encoding phase of an EM task in different aging stages. Two groups of adults were compared: middle-aged (age ranging from 50 to 64 years) vs. older (65 to 81 years) adults. Behavioral and ERP measures were collected in the recognition phase, yielding neural evidence about how tDCS affects the different processes involved in EM retrieval. We expected to find different neural and behavioral effects between age groups. Elucidating the behavioral and neural effects induced by tDCS along the healthy aging continuum may help the development of effective tDCS protocols specifically tailored to the aging brain, and to age-related neurodegeneration.

The only two significant differences among cognitive and demographical variables of the two groups were age (p < 0.001) and years of education (p < 0.01), with middle-aged adults being younger and showing a higher degree of education (Table 1).

Participants were unable to discriminate between the two tDCS conditions (anodal, sham). Over 36 participants, 25 responded they received “Real stimulation” in both sessions, 3 responded that they received “Sham” stimulation in both sessions, 7 selected “I do not know.” Only one participant was able to correctly deduct the stimulation received in both sessions.

Separated analysis performed on data collected in middle-aged and older adults showed no differences between anodal and sham tDCS on irritation, burning, heat, iron taste and fatigue sensations (middle-aged all p > 0.26; older adults all p > 0.18). Therefore, atDCS was indistinguishable from the sham tDCS condition in all the sensations; hence, the blinding between the two stimulation conditions reduced experimental biases on participants’ expectations (see Supplementary material for the data).

In the present study, we investigated the behavioral and neural effects of atDCS applied over the left DLPFC during the encoding phase of an EM Old-New recognition task in healthy middle-aged and older adults. Behavioral results showed that age modulated the effect of atDCS during the encoding on EM abilities, with an opposite pattern of results in the two age groups: whereas older adults improved their recognition performance, middle-aged participants showed a worsening in performance. This is the first study to highlight such an age-related difference in atDCS outcomes regarding memory tasks.

Present results on how tDCS applied during the encoding phase of an EM task differently affected the behavioral indexes in the two age groups may help to disentangle the specific mechanisms of action of tDCS upon EM processes. Indeed, a better/worse recognition performance can rely either on a more/less efficient encoding of target images (Old stimuli) or on a reduction/increase of interference of distractor stimuli (New stimuli) during the recognition phase. The present results showed that Hits (Old stimuli correctly judged as old) were enhanced by 2.50% and reduced by 2.41% after atDCS (compared to sham) in older and middle-aged adults, respectively. In contrast, CR (New stimuli correctly judged as new) were not affected by atDCS. Thus, we hypothesize that prefrontal atDCS affected recognition memory performance by strengthening (or decreasing) the memory traces of old to-be-remembered images (those presented during the delivery of tDCS) rather than by acting upon the response to distractor stimuli (New items) that have not been presented during the encoding phase. These findings appear consistent with an “online,” within-session effect of atDCS, rather than an “offline” post-tDCS session improvement (Sandrini et al., 2019), emphasizing that task component and age, not only tDCS, are relevant elements in determining the final effect.

Electrophysiological results showed that the behavioral deterioration observed in the group of middle-aged adults was mirrored with a modulation of specific EM neural correlates. Middle-aged adults showed a significant reduction in the left hemisphere EF and LF components after atDCS but not after sham tDCS. In contrast, the behavioral improvement observed in the group of older adults was not associated with any statistically significant modification of the electrophysiological components. To date, no study has investigated the effects of tDCS on these ERP signatures of EM in older adults.

It is widely held that recognition memory is supported by two functionally distinct processes, namely, familiarity and recollection (Friedman, 2013). The present ERP results showed that the first EM component affected by atDCS in the group of middle-aged adults was the EF effect, which represents a putative mechanism indexing familiarity-based recognition processes (Curran, 2000; Rugg and Curran, 2007). Through the reduction in EF amplitude in middle-aged adults, atDCS might have interfered with mnemonic processes relying on familiarity (i.e., fast and relatively automatic processes based on feeling that an event is old or new in the absence of confirmatory contextual information; Daselaar et al., 2006), causing a reduction in recognition efficiency in this age group.

Together with familiarity process modulation, the ERP results showed that atDCS also induced a reduction in the LF effect amplitude in middle-aged adults only in the stimulated left DLPFC. This later positive component is usually associated with post-decisional monitoring aspects of mnemonic processes more linked to executive functioning (Friedman and Johnson, 2000; Rugg, 2004; Hayama et al., 2008). However, atDCS over the left DLPFC did not induce any modulation of the recollection-based mnemonic processes, which are indexed by the parietal EM.

To our knowledge, there is only one other study that analyzed the effects of atDCS on ERP components elicited in an EM task (Lu et al., 2015). This study found modulations produced by atDCS in healthy young participants in the N400 and P600 ERP components, observable in the direct traces in a verbal EM task. The amplitude of N400 was enhanced in frontal regions and decreased in parietal regions after atDCS, while P600 amplitude was enhanced in parietal locations. However, the author do not discuss the functional significance of these modulations, nor the possible modulations over the old-new effects.

Interestingly, significant ERP amplitude reductions were restricted to the stimulated region (i.e., left prefrontal cortex) in the present results. The spatial specificity of this effect also aligns well with the hypothesized mechanism of action of atDCS on the behavioral measures (see above). In fact, there is evidence indicating that tDCS modulates neuro-metabolite levels at the site of stimulation and that this translates into alterations in behavioral outcomes (Chhabra et al., 2021). Notably, the “online” obtained effect on encoding processes is in accordance with the HERA (Hemispheric Encoding Retrieval Asymmetry) model, which postulates that the left DLPFC is crucially involved in the encoding of memory contents, whereas the right DLPFC is crucial for retrieval (Tulving et al., 1994; Nyberg et al., 1996; Habib et al., 2003; Manenti et al., 2012).

The opposite pattern of tDCS effects in middle-aged and older adults may point toward differential functionality of the system, highlighting the importance of the interaction of brain stimulation with brain activity in defining the behavioral outcome (Miniussi et al., 2010, 2013). Among the possible explanations put forward in the literature to unfold the inter-individual differences in tDCS efficacy, the present results may fit with previous studies supporting the hypothesis that tDCS gains are restricted to higher task demands (e.g., Li et al., 2015). Accordingly, older adults might experience a higher cognitive load than middle-aged adults when performing the same EM task and thus might benefit from atDCS-induced effects. However, cognitive load was not measured in this study, and no between-group difference in recognition performance was observed between groups after sham stimulation. Previous studies also suggested graded beneficial effects of tDCS from healthy young adults to physiological aging and then to pathological aging, with low cognitive performance groups benefitting more from the stimulation and with possible deleterious outcomes in healthy young adults who do not present a need for stimulation (Hsu et al., 2015). Although this framework appears promising for the interpretation of the present results, again, we did not observe any differences between middle-aged and older adults in cognitive functioning.

Another model that has been proposed refers to the “stochastic resonance,” which states that the same amount of stimulation may induce different ratios of signal-to-noise increase depending on the subject’s activation state (Miniussi et al., 2013; Fertonani and Miniussi, 2017). Consequently, the distinct neural activation that older adults might need to execute the task might have represented the optimal state to shape the beneficial behavioral outcome. Furthermore, it has been proposed that the leeway for tDCS modulation depends on the dissociation of the neural system from a previous optimal state. Therefore, whereas middle-aged adults might function close to their homoeostatic optimum (with a relatively small margin for tDCS improvement), older adults might shift away from it, possibly making their brain more amenable to tDCS-induced gains (Habich et al., 2020a,b).

However, in the absence of imaging data characterizing the neural state of the two groups of participants, it is difficult to draw a broader conclusion about the neural basis underlying the opposite tDCS effects.

Although the present results appear promising, some issues need to be addressed. First, the limitations of the cross-sectional designs need to be considered when interpreting the present results. A longitudinal study assessing the within-subjects progression along the aging lifespan is needed to verify how the impact of tDCS on EM changes as the aging process advances. Another limitation that needs to be addressed regards the lack of a control condition, such a control target site and/or different intensities. This type of control conditions are important to ensure that changes in memory performance are specific to the protocol adopted. However, our study did not aimed at investigating the optimal tDCS protocol to induce memory improvements but to assess whether the same protocol applied in different aging stages induced differential effects. We want to emphasize that it remains to be clarified whether our results generalize to other tDCS electrode configurations and stimulation parameters, highlighting the importance and non-triviality of replicating and extending our observations while varying methodological factors.

In our study, we did not have MRI scans of the participants, thus impeding us from controlling for whether different levels of cortical atrophy between middle-aged and older adults may have been associated with different electric field characteristics.

Our experimental design did not include a baseline assessment in the EM task, like in several other paradigms where the main aim was to assess tDCS effects on memory in two different groups (e.g., Manenti et al., 2013; Brambilla et al., 2015; Fiori et al., 2017; Leach et al., 2019). The inclusion of a baseline would have augmented the length of experimental sessions, with possible detrimental effects, e.g., increasing fatigue level and/or decreasing participants’ adherence to the project. Importantly, we did not find differences between groups in any of the behavioral and ERP variables in the sham condition nor in any of the neuropsychological measures at baseline. Moreover, we ensured that experimental conditions were exactly the same in both experimental sessions except for the tDCS condition. Hence, although group differences at baseline in this specific EM task cannot be completely ruled out, it is likely that behavioral and neural differences are due to the application of atDCS.

Here, we provide evidence for an age-dependent effect of atDCS over the DLPFC on EM and its underlying electrophysiological substrates, indicating different and opposing modulatory patterns along the healthy aging continuum. These results contribute to a better understanding of the differential age effects of tDCS, giving support to the notion that tDCS combined with a task can be used to improve cognition only under given conditions. Because of altered neuroplasticity and network dynamics with age, a detailed investigation of older adults covering a wide age range is of paramount importance. Such an investigation would both elucidate the complexity of tDCS effects at the neurophysiological level and help develop more individually tailored interventional protocols in healthy populations and in age-related diseases, such as Alzheimer’s disease and dementia.

The original contributions presented in the study are publicly available. This data can be found here: https://gin.g-node.org/MartaBortoletto/Bagattini_Cid-Fernandex_et_al_2023.

The studies involving human participants were reviewed and approved by Ethics Committee of the IRCCS Centro San Giovanni di Dio Fatebenefratelli (Via Pilastroni, 4 Brescia – Italy). The patients/participants provided their written informed consent to participate in this study.

CB: conceptualization, methodology, investigation, formal analysis, data curation, writing – original draft, writing – review and editing, visualization, and project administration. SC-F: methodology, investigation, software, formal analysis, writing – original draft, writing – review and editing, and visualization. MBu: formal analysis and writing – review and editing. CM: conceptualization, methodology, writing – review and editing, supervision, and funding acquisition. MBo: conceptualization, methodology, writing – review and editing, project administration, and supervision. All authors contributed to the article and approved the submitted version.

This study was funded by the Italian Ministry of Health Ricerca Finalizzata to CM (RF-2013-02356444) and Ricerca Corrente. SC-F was founded by Galician Postdoctoral Grants Plan (ED481B 2016/078-0, Xunta de Galicia).

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.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.