Practice improves older adults’ attentional control and prospective memory more than HD-tDCS: a randomized controlled trial

Abstract

Frontal and parietal brain regions are involved in attentional control and prospective memory. It is debated, however, whether increased or decreased activity in those regions is beneficial for older adults’ task performance. We therefore aimed to systematically modulate activity in those regions using high-definition transcranial direct current stimulation. We included n = 106 healthy adults (60–75 years old, 58% female) in a randomized, double-blind, and sham-controlled study. We evaluated task performance twice in the laboratory and at home and additionally assessed heart rates. Participants received cathodal, anodal, or sham stimulation of the left or right inferior frontal lobe, or the right superior parietal lobe (1 mA for 20 min). Performance improved at visit two in laboratory tasks but declined in at-home tasks. Stimulation did not modulate performance change in laboratory tasks but prevented decline in at home-tasks. Heart rates increased at visit two but only when right inferior frontal lobe activity was inhibited. Repeating a task seems more beneficial than stimulation for laboratory tasks. This might be different for at-home tasks. Inhibiting right frontal brain function increases heart rates, possibly due to a modulation of the frontal-vagal brain-heart axis.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-024-74029-9.

Introduction

Remembering to do something in the future has been referred to as prospective memory¹. Prospective memory contains a prospective component (i.e., remembering that something has to be done) and a retrospective component (i.e., remembering what to do and when^1,2). The retrospective component depends on long-term memory while the prospective component relies on attentional control^1,3,4. Attentional control declines with age, which may explain why older adults struggle to remember prospective intentions^5–8. At a neural level, prospective memory and attentional control seem to be linked as well, as for example the inferior frontal lobe or the superior parietal lobe were active in both tasks^9–12. In ageing, though, it is not fully understood whether increased or decreased activity in those regions is beneficial for task performance^12–18. Studies in younger adults suggest that enhancing activity in the right inferior frontal lobe or the right superior parietal lobe using non-invasive brain stimulation improved attention or attentional control^19–22. A similar effect was found in older adults when modulating right inferior frontal lobe function²³. These studies rely on behavioural outcomes to draw conclusions on the effects the stimulation had at the neuronal network level. Direct measurements of brain function outcomes following non-invasive brain stimulation in humans show that anodal tDCS enhanced brain activity while cathodal tDCS decreased it^24–26.

For prospective memory, inhibiting or enhancing left (but not right) superior parietal lobe function modulated performance in younger adults—inhibition led to better accuracy while facilitation led to slower responses^27,28. Using fMRI, more activity in the left inferior frontal lobe was found to be associated with faster prospective memory responses in younger adults¹⁵. It remains incompletely understood whether this is also the case in older adults. This is an important question because, if so, this will help gain insight into the mechanisms of prospective memory function in older adults.

We therefore aimed to systematically modulate activity in the right superior parietal lobe, the right inferior frontal lobe, or the left inferior frontal lobe in healthy older adults using high-definition transcranial direct current stimulation (HD-tDCS). In contrast to conventional tDCS, which leads to broad and diffuse stimulation of target regions and their surrounding (or even distal) areas, HD-tDCS leads to very focal stimulation of a targeted region. It has been reported that repeating a cognitive task leads to better task performance^29–32. Stimulation might further enhance these practice effects (i.e., learning effects). Hence, we assessed performance in all tasks twice (one week apart) and applied stimulation at the second visit. In line with the pre-registered protocol of this study, we hypothesized that stimulation of the right inferior frontal lobe will lead to changes in prospective memory performance and attentional control. Whether cathodal or anodal stimulation will enhance performance is not clear yet, since previous fMRI studies were inconsistent about activity changes in older adults. Further, we expected that anodal stimulation of the left inferior frontal cortex will lead to faster responses to prospective memory stimuli, while cathodal stimulation of the same area will prolong reaction times. We hypothesized that attentional control will not be affected by stimulation of the left inferior frontal lobe. Finally, we expected that anodal stimulation of the right superior parietal cortex will lead to better prospective memory and attentional control performance, while cathodal stimulation will have detrimental effects on both cognitive processes.

Materials and methods

The study was done in accordance with the SPIRIT guidelines³³. We pre-registered the study (clinicalTrials.gov identifier: NCT04882527; date of first registration: 12/05/2021) and followed the CONSORT guidelines³⁴. A detailed description of the study procedures and statistical analysis can be found in our study protocol³⁵.

Study design and randomization

We used a randomized (allocation ratio 1:1), double-blind, sham-controlled, and parallel-group design. Participants were stratified by age and sex (self-reported) and received anodal or cathodal HD-tDCS of the right inferior frontal lobe, the left inferior frontal lobe, or the right superior parietal lobe or sham. Randomization was done by a person not involved in data collection (JP) using in-house Matlab scripts. Similar to the real tDCS stimulation group, in the sham group, the site of sham stimulation was randomly allocated. Participants and investigators (NS, RM) were unaware of the experimental condition.

Participants

We included n = 106 healthy older adults (mean age 67.7 ± 4.1 years; 58.5% female; Table 1; Fig. 1). The required sample size was based on an a-priori G*Power analysis as outlined in our study protocol. We have used an effect size of η = 0.32 from a previous study²⁸and the options α = 0.001, β = 0.95, ANOVA with main effects and interactions, 7 groups). We started the trial in June 2021 and ended data acquisition in October 2023 (we recruited participants from June 2021 until June 2023). All participants were fluent in German, right-handers (or ambidextrous), non-smokers, and had normal or corrected-to-normal vision. They reported no history of severe psychiatric or neurological disorders or injuries, no current or lifetime seizures, no alcohol or drug abuse, and no use of medications that interfere with cognition. In addition, they did not report dermatosis, metal implants in the head area, brain or cardiac pacemakers, or stents. None of the participants showed signs of cognitive impairment (all Montreal Cognitive Assessment scores were ≥ 23³⁶) or clinically relevant depressive symptoms (all Geriatric Depression Score were ≤ 6³⁷). All participants provided written informed consent before testing. The Ethics Committee of the Canton of Bern approved the study (ID: 2019-01599, date: 16.10.2019), which was conducted according to the Declaration of Helsinki. All data were acquired at the University Hospital of Old Age Psychiatry and Psychotherapy in Bern, Switzerland.

Table 1.

Sociodemographic characteristics of the sample (mean and standard deviations).

	Anodal			Cathodal			Sham (–)
	FC5	FC6	P4	FC5	FC6	P4
N (number female)	15 (9)	15 (9)	15 (8)	15 (9)	16 (9)	15 (9)	15 (8)
Age in years	67.1 (4.3)	68.6 (4.4)	67.8 (4.1)	67.9 (3.6)	68.1 (3.0)	68.2 (5.1)	66.1 (4.1)
Education in years	17.3 (2.9)	17.7 (2.4)	16.7 (2.6)	17.5 (2.6)	16.2 (2.7)	16.2 (2.6)	16.7 (3.1)
Verbal IQ	115.7 (9.4)	116.1 (7.4)	110.0 (8.9)	116.6 (7.5)	110.8 (8.7)	115.2 (7.2)	110.6 (7.6)
MoCA (0–30)	26.7 (1.9)	27.0 (1.6)	26.9 (1.8)	27.7 (1.8)	26.9 (2.3)	26.3 (2.2)	26.9 (2.1)
GDS (0–15)	0.4 (0.5)	0.8 (0.9)	0.7 (0.9)	1.3 (1.5)	1.4 (1.6)	0.5 (0.6)	0.7 (0.9)

For MoCA and verbal IQ, higher scores indicate better performance. For the GDS, lower scores indicate fewer depressive symptoms.

FC5 left inferior frontal lobe, FC6 right inferior frontal lobe, P4  right superior parietal lobe, IQ Intelligence quotient, MoCA  montreal cognitive assessment, GDS geriatric depression scale.

Fig. 1.

Participant flow diagram. Flow diagram of the progress through the phases of a 2-group parallel randomized controlled trial.

Study procedures

All participants were tested twice to examine whether learning effects can be modulated by stimulation. In brief, at visit one, we screened participants for cognitive impairment or depressive symptoms, and they rated their current mood. Afterwards, we tested their attentional control and laboratory prospective memory while assessing their heart rate using a wearable device. Finally, we tested their naturalistic prospective memory. At visit two (approximately one week later), we used the same tasks and the same wearable device while participants received HD-tDCS. Stimulation started at the beginning of the attentional control task and ended 20 min later, when the laboratory prospective memory task was completed. Then, naturalistic prospective memory tasks followed just as at visit one. At the end of visit two, any side effects³⁸, as well as the participants’ perception of the stimulation condition were captured.

High-definition transcranial direct current stimulation (HD-tdcs)

We used 4 × 1 ring electrodes for HD-tDCS (Fig. 2; Soterix Medical, NY, USA). The position of the central electrode was located according to the 10–20 EEG system: for right or left inferior frontal lobe stimulation, we placed the central electrode over FC6 or FC5 and for the right superior parietal lobe over P4 (Fig. 2). The adjoining electrodes were placed in a ring-shape approximately 3.5 cm away from the central electrode. Each electrode had a diameter of 1.2 cm. We used the same electrode positions for sham or real stimulation. Real tDCS consisted of 30 s ramp-up, followed by constant current at 1 mA for 20 min and ramp-down for 30 s. For sham stimulation, the current was ramped up to a current of 1 mA just as for real tDCS but was immediately ramped down again (at the beginning and the end of 20 min). This sham procedure produces similar sensations as real stimulation but without exerting any stimulation effects^39,40. To prevent the electrodes from heavy and irregular abrasion, we regularly changed the central electrode^41,42. In addition, each set of electrodes was used for 35 participants (i.e., each electrode was used seven times as central and 28 times as surrounding electrode). We quantified the amount of added gel (Signa Gel) used in all participants and checked lead quality of the electrodes after mounting, right before the start of the stimulation, and at the end (i.e., before removing the electrodes).

Fig. 2.

Electrode positions and simulations of current flow through the brain. Electrodes were mounted over the right inferior frontal lobe (FC6; A), the left inferior frontal lobe (FC5; B), or the right superior parietal lobe (P4; C). Positive values (red) indicate an expected increase, negative values (blue) an expected decrease in excitability of neurons. Simulations were created with SimNIBS (version 3.2.6¹⁰⁰). Note: From Schmidt et al.³⁵. CC BY-NC.

Attention network task

We presented the attention network task^43,44 using PsychoPy3 (version 2021.1.3). In brief, participants pressed one of two buttons to indicate the direction of a central arrow. The arrow was flanked by other arrows pointing in the same direction (i.e., congruent condition), the opposite direction (i.e., incongruent condition), or it was flanked by squares (i.e., neutral condition). The stimuli appeared either above or below a fixation cross. In some trials, a cue preceded the appearance of stimuli, informing participants of the timing and, occasionally, the location of following arrows. It is possible to calculate three aspects of attention with this task: Alerting, orienting, and executive control (i.e., ‘controlling’). We were primarily interested in executive control (i.e., the ability to ignore distracting and task-irrelevant information). Therefore, we only report the difference in response times between congruent and incongruent flanker trials, with lower values indicating better performance. We excluded one participant from statistical analysis due to poor task performance (< 50% correct responses).

Laboratory prospective memory task

An event-based prospective memory task followed the attentional network task⁴⁵ and was presented using PsychoPy3 (v2021.1.3). In brief, the task consisted of an ongoing task and a prospective memory task. For the ongoing task, participants had to indicate by button presses whether a pair of words belonged to the same semantic category or not. For the prospective memory part, they occasionally saw a string of letters (either ‘cccc’ or ‘vvvv’) instead of a word pair. For these stimuli, they were instructed to memorize both the letter in the string and its colour. Whenever one of the subsequent word pairs appeared in the same colour as the previous string, they had to press the corresponding letter key (‘v’ or ‘c’). Since we were primarily interested in prospective memory, we report task accuracy (i.e., correctly remembered ‘c’ or ‘v’ button presses) and reaction times of correctly remembered button presses. We excluded three participants from statistical analysis because they did not remember any prospective intention (2 participants) or due to a very high number of false alarms (1 participant pressed the prospective memory button 28 times when the instruction was to press the 1-back button, suggesting wrong understanding of the task; the mean of other participants was 2.9, median was 2, in keeping with an occasional error).

Test battery for attentional performance

Next, we tested attentional control using three different tasks from the test battery for attentional performance (TAP⁴⁶). In brief, in a go/no-go task they had to respond to go trials while ignoring no-go trials. Then, a flexibility task followed, were they had to respond to switching targets (i.e., numbers or letters). Finally, they completed a divided attention task that required simultaneous visual and auditory processing. Since we were primarily interested in attentional control, we report false alarms in the go/no-go task, response times in the flexibility task, and missed targets in the divided attention task as well as response times of correct answers. We excluded two participants from statistical analysis of the go/no-go task as they probably confused go- and no-go stimuli, resulting in > 50% error rate.

Naturalistic prospective memory tasks

Finally, we tested naturalistic prospective memory using an adapted version of the Royal Prince Alfred Prospective Memory Task ⁴⁷). In brief, participants had to remember two short-term and two long-term prospective intentions. The short-term intentions had to be remembered during each session, one after five minutes (i.e., time-based task) and another when the last task was started (i.e., event-based task). The two long-term intentions had to be remembered when arriving at home (i.e., event-based task) and when 5 days had passed (i.e., time-based task). We asked participants not to use any external memory aids. We rated performance on a scale ranging from 0 (late and incorrect retrieval) to 3 (timely and correct retrieval) and calculated an overall score ranging from 0 to 12 per session.

Heart rate

At both visit one and two, participants put on a wearable device (Polar Verity Sense by Polar Electro Europe AG, polar.ch) on their left arm with an elastic strap to monitor their heart rate (measured in beats per minute). We collected data for 24 min, covering the attention network task and the prospective memory task (plus two minutes before and after). At visit two, this time frame covered the stimulation time (plus two minutes before and after). We excluded any data points outside 50−110 bpm (i.e., ectopic beats). We calculated a baseline heart rate for each participant using the mean heart rate within the first two minutes of assessment. This baseline value was subtracted from each following data point to calculate a ‘baseline corrected’ heart rate for the remaining 22 min⁴⁸. We then computed a mean corrected heart rate in 5-second intervals to capture rapid changes, providing a fine-grained temporal resolution, and to reduce computation time of generalized equation models. Four participants with more than 20% missing values were excluded from this analysis.

Current mood and affect

We assessed current mood using the Profile of Mood States (POMS^49,50) and current affect using the Positive and Negative Affect Scale (PANAS⁵¹) at the beginning and the end of each visit. We used the POMS total score and as well as positive and negative affect derived from PANAS for statistical analysis.

Statistical analysis

We outlined the statistical analysis in detail in our study protocol. In brief, we first determined whether inhibiting or enhancing activity in the left or right inferior frontal lobe or the right superior parietal lobe modulated attentional control or prospective memory. We used mixed ANOVAs with session (first vs. second) as within-subject factor and group (anodal or cathodal stimulation of right inferior frontal lobe, left inferior frontal lobe, right superior parietal lobe, or sham) as between-subject factor and prospective memory task accuracy and response times, or false alarms in the go/no-go task, missed targets in the divided attention task, executive control and response times as dependent variable. Next, we tested whether stimulation had any effect on mood or heart rate (exploratory analysis). We used mixed ANOVAs with time-point (beginning vs. end of visit) as within-subject variable and group as between-subject variable. These ANOVAs were calculated for visit one and two separately. For heart rate, we used generalized equations modelling^52–54to account for hierarchical dependencies and autoregressive structure as generalized equations modelling is more convenient than mixed models for group differences in repeated designs⁵⁴. In case of significant stimulation effects, we correlated each participants’ heart rate slope (i.e., change in heart rate over the visit calculated with a linear regression per participant) with their cognitive performance or change in mood (exploratory analysis). To examine whether the amount of conductive gel was similar in all groups, we conducted an ANOVA with group as a between-subject factor. Finally, we evaluated blinding success using Chi-square-tests and tested, whether there were any group differences in side effects using one-way multivariate ANOVA with group as between-subject factor.

We used R (version 4.2.1) with RStudio (version 2022.02.3) and the packages WRS2⁵⁵, and geepack⁵⁶. We used R or GraphPad Prism (version 9.0.0; USA) for visualisation of the results. We considered p < 0.05 for statistical significance (p < 0.10 was considered for interpretation in some cases). Given a main effect, post-hoc tests of primary outcome measures were corrected for multiple-comparisons using Bonferroni-Holms’ procedure. We used robust alternatives such as ANOVAs on trimmed means or Kendalls’ τ for correlations when normality or homogeneity assumptions were violated or in case of extreme outliers.

Results

All groups were comparable regarding age, sex-ratio, and education (Table 1; all p > .69). In addition, MoCA scores, depression scores, verbal intelligence, hours of sleep, motivation, concentration, or wakefulness were similar in all groups (all p > .08).

Participants reported only few and mild side effects (Table S3). The two most reported side effects were tingling and burning sensations. 45% reported mild and 7% reported moderate tingling sensations. 34% reported mild, 9% reported moderate, and another 3% reported strong burning sensations. The occurrence of side effects was similar in all groups (F_{(6, 57.0)} = 1.26, p = 0.10; n = 106). When asked whether they thought they had sham or real tDCS, participants’ (χ² = 0.289, p = 0.79) and experimenters’ (χ² = 1.809, p = 0.25) responses were at chance level. The impedance after mounting the electrodes (n = 105, p = 0.52), before starting the stimulation (n = 106, p = 0.34), and at the end (n = 104, p = 0.09) was comparable in all groups. Contact quality was rated moderate or good during all stimulations. We used a comparable amount of gel for the electrode montage in all participants (~ 9.1 g on average, p = 0.67).

All participants improved in attentional control and laboratory prospective memory-stimulation had no additional effect

All participants remembered laboratory prospective intentions more accurately (F_{(1, 47)} = 37.12, p < 0.001; n = 103), and faster (F_{(1, 48)} = 5.05, p = 0.03; n = 96) at visit two than at visit one (Table 2). We did not find an additional effect of stimulation (all p > 0.24; see Supplement S1). We found similar results for attentional control and executive functions (Table 2): Controlling (F_{(1, 44)} = 23.89, p < 0.001; n = 105) improved significantly and participants had fewer false alarms (F_{(1, 43)} = 17.02, p < 0.001; n = 104), fewer missed dual task responses (F_{(1, 34)} = 34.78, p < 0.001; n = 106), and they responded faster to dual task stimuli (F_{(1, 52)} = 6.95, p = 0.01; n = 106) at visit two compared to visit one. Again, we did not find any additional stimulation effect (all p > 0.19, see Supplement S1).

Table 2.

Task performance (mean and standard deviations) in healthy older adults receiving high-definition transcranial direct current stimulation of frontal or parietal brain regions (or sham) at visit two.

	Visit one	Visit two
Prospective memory task, remembered intentions (0–12)	5.7 (3.1)	8.2 (2.9)
Prospective memory task, response time (in s)	1.4 (0.4)	1.3 (0.3)
Controlling (in s)	0.2 (0.1)	0.1 (0.1)
Go/No-go (number of false alarms)	1.7 (2.2)	0.8 (1.2)
Flexibility response time (in s)	1.0 (0.4)	0.8 (0.3)
Dual task (number of missed responses)	2.3 (3.1)	1.9 (2.4)
Dual task response time (in s)	0.8 (0.1)	0.7 (0.1)

For the variable ‘controlling’, lower values indicate better task performance.

Change in naturalistic prospective memory was differentially modulated by stimulation

Participants remembered naturalistic intentions better at visit one than visit two (F_{(1, 48.3)} = 4.55, p = 0.04; n = 104; Fig. 3). Hence, naturalistic prospective memory declined at visit two. We found a significant interaction between visit and stimulation group (F_{(6, 27.7)} = 2.69, p = 0.04). Post-hoc tests revealed that with anodal stimulation of the right inferior frontal lobe naturalistic prospective memory did not decline, while in the group receiving cathodal stimulation of the right superior parietal lobe the decline was even more pronounced with stimulation. However, this effect was only at trend level significant after controlling for multiple comparisons (p = 0.08). In addition, there was no significant difference between any stimulation group and sham.

Fig. 3.

Change (visit two minus visit one) in older adults’ naturalistic prospective memory with frontal or parietal transcranial direct current stimulation or sham. Electrodes were mounted over the right inferior frontal lobe (FC6), the left inferior frontal lobe (FC5), or the right superior parietal lobe (P4). Positive values indicate an improvement, negative values indicate worsening of performance.

Stimulation had no significant effect on mood or affect

Participants reported higher negative mood at the end of visit one (F_{(1, 35.9)} = 7.96, p = 0.007). In contrast, their affect was less negative at the end of visit two (F_{(1, 37.6)} = 30.8, p < 0.001). Stimulation had no significant effect on any of those ratings (all p > 0.27, see Supplement S1).

Heart rate decreased in most participants, while it increased when inhibiting right inferior frontal lobe function

At visit one, the heart rate of participants decreased over time (χ²₍₁₎ = 23.69, p < .001; n = 102), with no difference between groups (p = 0.72). At visit two, we found a significant interaction between stimulation group and time-point (χ²₍₆₎ = 22.90, p = 0.008). Post-hoc tests contrasting change in heart rate as a function of time and stimulation group revealed that cathodal stimulation of the right inferior frontal lobe slightly increased the heart rate over time while the heart rate decreased in those receiving anodal (β = − 0.029, SE = 0.008, z = − 3.83, p = 0.002) or cathodal stimulation of the left inferior frontal lobe (β = − 0.019, SE = 0.006, z = − 3.26, p = 0.02), or sham (β = − 0.017, SE = 0.005, z = 3.18, p = 0.03, Fig. 4). Only cathodal stimulation of the left inferior frontal lobe resulted in a slight increase of the heart rate (i.e., a positive slope) when comparing beginning and end of session (Fig. 4).

Fig. 4.

Change in heart rate (beats per minute, bpm) between beginning and end of visit two in older adults receiving frontal or parietal high-definition transcranial direct current stimulation or sham. Electrodes were mounted over the right inferior frontal lobe (FC6), the left inferior frontal lobe (FC5), or the right superior parietal lobe (P4). Positive values indicate an increased heart rate at the end of visit two, negative values indicate a reduced heart rate.

Change in heart rate was correlated with change in mood and attentional control

We found significant correlations between heart rate slopes (i.e., individual changes in heart rate over time) and change in negative affect (Kendall’s τ = − 0.14, p = 0.044) as well as with attentional control (derived from the attention network task, Kendall’s τ = 0.18, p = 0.008). These results indicate that an increasing heart rate was associated with less negative affect but worse attentional control.

Discussion

In this study, we tested whether anodal or cathodal tDCS over frontal or parietal areas influenced attentional control or prospective memory in older adults. Prospective memory was tested both in the laboratory and at home (i.e., naturalistic task). To examine whether stimulation modulated learning effects, participants performed the same tasks twice and we stimulated at visit two. We found learning effects with repeated assessment but no additional effect of stimulation for laboratory tasks. For tasks performed at home, we found worsening of performance at visit two and a trend indicating an even more pronounced worsening when inhibiting right superior parietal lobe function. However, this trend was only visible in comparison to enhancing right inferior frontal lobe function and not in comparison to sham. In addition, our results indicate that inhibiting right inferior frontal lobe function increased the heart rate. Regardless of stimulation group, a decrease in heart rate was associated with better mood and better attentional control. The results of our study are important in two aspects. First, learning effects seem to be stronger than stimulation effects for laboratory tasks. In contrast, in naturalistic tasks, performance worsened over time, and worsening was slightly more pronounced with inhibition of the right superior parietal lobe (although the effect was at trend level significant). This indicates that inhibiting parietal lobe function may negatively influence performance in tasks that are closer to older adults’ daily life. Second, inhibiting frontal brain function increased the heart rate of participants and changes in heart rate were related to mood and attentional control. However, these results should be interpreted with caution as they were exploratory.

Inhibiting right parietal brain function may be harmful for naturalistic prospective memory

Our results have important implications. All participants improved in attentional control and laboratory prospective memory with practice, supporting previous study findings^{29–32,57,58}. We found no evidence to suggest that enhancing or inhibiting frontal or parietal brain function improved laboratory task performance beyond learning. This contradicts previous findings in which enhanced right inferior frontal lobe function improved attentional control in healthy younger or older adults^19,23. These studies, however, did not test learning effects so it could be that enhancing right frontal lobe function is only beneficial for attentional control when there are no substantial learning effects. Support for this comes from previous studies that focused on episodic memory. We have repeatedly shown that tDCS enhanced episodic memory in younger adults^59–61. In these studies, we saw this effect with tDCS when participants were engaged in a novel task. In contrast, we did not find a similar effect in younger participants when they performed a slightly adapted (yet similar) task on a separate day while being stimulated⁶². Hence, the modulating effect of tDCS may be more pronounced when a task is novel and there were not yet substantial learning effects. We found similar results when modulating left frontal lobe function and right superior parietal lobe function. Again, we did not find evidence to suggest that stimulation had any additional effects beyond learning for laboratory tasks. This would indicate that if we want to improve attentional control or laboratory prospective memory, repeating the task may be more efficient than using stimulation.

For naturalistic prospective memory, however, participants became worse at visit two. It could be that the appointment for visit two helped them remember their intention after visit one. This would be in line with studies reporting improved prospective memory performance of older adults when reminders were provided^63,64. Since there was no reminder after visit two, their performance became worse. With modulation of right superior parietal lobe function, performance was even worse. This confirms our hypotheses as we anticipated that inhibiting right superior parietal brain function would decrease prospective memory performance. It indicates that normal right superior parietal lobe function is important for normal naturalistic prospective memory performance. When enhancing right frontal lobe function, in contrast, the drop in performance was slightly less pronounced. The right inferior frontal lobe seems to be important for the suppression of interfering memories as reduced brain function in that area led to worse memory or enhanced interference⁶⁵. Having control over interfering memories may be particularly important for prospective memory since different intentions have to be kept in mind at the same time. In our study, participants were asked to remember very similar intentions (e.g., telling the experimenter what they had for breakfast or what they wanted to cook for dinner). This may have confused participants about what to do and when. Interference may have been particularly high at visit two since old intentions from visit one may still have been present in their memory. Possibly, enhancing right frontal lobe function helped them control those interferences, thereby enhancing retrieval of naturalistic prospective intentions.

Inhibiting right inferior frontal lobe function increased the heart rate

As an exploratory analysis, we examined how participants heart rate changed at visit 1 (beginning to end of session; no stimulation) and visit 2 (beginning to end of session; with stimulation). In general, one would expect a decrease over time as participants get used to the testing situation. This was indeed the case at visit 1, were heart rates significantly decreased in all participants. At visit 2, however, participants’ heart rate did not decrease when we inhibited inferior frontal lobe function. This indicates that participants in this group became less used to the testing situation (i.e., they were similarly aroused at the end as in the beginning). According to the frontal-vagal network theory⁶⁶ and the neurovisceral integration model⁶⁷, frontal brain regions (e.g., the dorsolateral prefrontal cortex) control activity of the vagus nerve, which is the direct autonomic connection to the heart. In addition, increased prefrontal brain activity was linked to a lower heart rate and increased heart rate variability^66,67. In younger adults, left and right prefrontal brain regions seem to modulate the heart rate. With age, the influence of the left hemisphere decreases, and only right prefrontal brain regions seem to modulate the heart rate⁶⁸. If inhibiting right inferior frontal brain function modulated the vagus nerve, our results are consistent with these findings and confirm previous non-invasive brain stimulation studies^69,70. We also found that an increased heart rate was associated with reduced attentional control. This indicates that participants performed worse if they became increasingly aroused during cognitive testing. On the other hand, an increased heart rate was associated with less negative affect, indicating that becoming more aroused is not necessarily perceived as negative as it might lead to more positive emotional states. Heart rate, mood, and cognitive performance seem to be interrelated^71–77. Better executive functions have been associated with increased heart rate variability^71,74, and lower heart rate after a cognitive task was linked to better memory⁷⁷. In contrast, older adults with high arousal had higher heart rates and poorer working memory⁷⁸, indicating that both too much and too little activity of the autonomic nervous system can reduce cognitive performance⁷¹. Luo et al.⁷³ tested daily valence and arousal in combination with a prospective memory task and heart rate in older adults. They found that high arousal due to negative affect impaired prospective memory, but only if heart rates were high at rest. It would be important to further investigate how frontal brain functions influence the connection between heart rate, arousal, mood, and cognition. One suggestion could be to conduct a study similar to Luo et al.⁷³ that also includes a naturalistic prospective memory task and that compares stimulation of the left inferior frontal lobe to stimulation of the right inferior frontal lobe.

Other explanations for stronger practice effects than stimulation effects in laboratory tasks

There are several other explanations why laboratory performance did not change with modulation of frontal or parietal brain function. First, the effects of learning were so substantial that participants reached a certain threshold and modulation of brain function could not improve performance any further. Second, brain activity during a task may influence stimulation outcome. Third, we did not use the most suitable stimulation parameters.

Regarding the first, we found substantial learning effects from visit one to visit two. Hence, one might argue that there was no further room for improvement with stimulation. This is supported by research showing that low performers benefit the most from tDCS^60,79–81. A related point could be that stimulation may lead to stronger effects in untrained than in trained individuals^82–84(our participants trained the tasks shortly and only at visit one). Support for this comes from previous studies focusing on episodic memory. We have repeatedly shown that tDCS enhanced episodic memory (though in younger adults)^59–61. In these studies, we saw this effect with tDCS when participants were engaged in a novel task. In contrast, we did not find this when they performed a slightly adapted (yet similar) task on a separate day while being stimulated⁶². Hence, the modulating effect of tDCS may be more pronounced when a task is novel and there are not yet substantial learning effects. Still, future studies may systematically manipulate when to stimulate to find out when and how participants can benefit from stimulation.

Regarding the second, the amount of brain activity during a task may have played a role (see e.g., Sebastian et al.¹²) as this seems to influence whether stimulation will lead to behavioural changes^81,85. Some participants may benefit from increased brain function (i.e., using anodal stimulation), others from reduced brain function (i.e., using cathodal stimulation), and still others will not benefit at all. Consequently, at a group level, effects of stimulation may not be visible⁸⁶. Closed-loop systems that measure brain activity (e.g., using EEG or fMRI) and that adapt stimulation based on that activity may be a way to overcome this problem⁸⁷. A more practical approach might be to use participant-specific neuroimaging data to determine brain regions to be targeted and to define the direction of stimulation^88–90.

Regarding the third, HD-tDCS enhances or inhibits brain functions very focally, while conventional tDCS influences broader brain networks⁹¹. Cognitive functions such as attention or memory rely on large networks rather than on one brain region alone. Therefore, enhancing or inhibiting those cognitive functions may have worked better with conventional rather than HD-tDCS.

In addition, current density or stimulation length may have influenced the results. On the other hand, no clear link between HD-tDCS effects and current intensity or stimulation length has been found so far, as only few studies are available^92,93. It has even been suggested that longer or stronger stimulation protocols do not necessarily lead to better cognitive performance⁹⁴. Using tDCS several times rather than once may be an alternative approach to enhance cognition^92,95,96. Another possibility may be to adjust the distance between central and surrounding electrodes. It has been found that larger distances lead to less focal but stronger and deeper electrical fields⁹⁷. In our study, we used a minimum distance of 3 to 3.5 cm. For future studies, distances ranging from 4 to 5 cm may achieve focal yet stronger/ deeper electrical fields.

Limitations

Our study may have several limitations. Although our sample size was relatively large, the individual intervention sample size was rather small. This may have influenced our findings. Second, localizing electrode positions may have been more precise had we used MRI-guided neuronavigation. Our study, however, followed a standardized protocol according to established guidelines⁹⁸. That is, we placed electrodes according to the 10–20 EEG system and an EEG cap kept the electrodes in place during the experiments that were all done by the same experimenters. Third, heart rate variability has been found to be associated with cognitive functions. Hence, it would have been more appropriate to use an electrocardiogram instead of a wearable device to assess change in heart rate.

Conclusion

Our study indicates that learning increases laboratory task performance, while modulation of frontal or parietal brain activity does not. For tasks performed at home, modulation of right frontal lobe function may slightly reduce forgetting, while inhibiting right parietal brain function may slightly enhance forgetting. Inhibiting right frontal brain function altered the heart rate, indicating that HD-tDCS can modulate the autonomic nervous system in older adults.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contributions

Conceptualisation: JP; Methodology: JP, NS, MK; Investigation: NS, RM, MMG; Analysis: JP, AH, NS; Visualisation: NS, JP; Funding acquisition: JP; Project administration: JP; Supervision: JP; Writing – original draft: NS, JP; Writing – review and editing: All authors.

Funding

JP has received funding from the Swiss National Science Foundation (grant numbers: 185105, 218252). The funding agency had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Data availability

The datasets used and/or analysed during the current study available from the corresponding author on reasonable request.

Declarations

Competing interests

The authors declare no competing interests.