Exploring the Mechanisms of Temporal Interference Stimulation in Enhancing Spatial Working Memory Performance through Frontoparietal Modulation

doi:10.21203/rs.3.rs-8187601/v1

Download PDF

Research Article

Exploring the Mechanisms of Temporal Interference Stimulation in Enhancing Spatial Working Memory Performance through Frontoparietal Modulation

https://doi.org/10.21203/rs.3.rs-8187601/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Temporal Interference Stimulation (TIS) has garnered attention for cognitive modulation; however, its mechanisms remain debated. This study introduces a novel protocol of theta-band TIS on spatial working memory (SWM) by targeting two regions (IPL, MFG) simultaneously, utilizing a change detection paradigm and fMRI to assess behavioral and neural changes. Behavioral results demonstrate TIS’s capacity to significantly enhance SWM performance by decreasing reaction times. Accordingly, neural activity was reduced in relevant regions during encoding and retention stages, correlating with better performance. Notably, functional connectivity in IPL and surrounding regions was enhanced during retention. These findings indicate that TIS functions to improve the processing efficiency of the neural network for SWM in that less neural resources were needed and therefore recruited in relevant regions, and the functional connections became stronger among these regions. Our results are against the notion that TIS taxes more neural resources to supplement a cognitive task.

temporal interference stimulation

spatial working memory

frontoparietal network

fMRI

As a novel neuromodulation technology, temporal interference stimulation (TIS), since its introduction [1], has garnered widespread attention due to its non-invasiveness, safety, and ability to target deep brain regions. These features promise broad applications in both research and clinical settings. However, studies on the modulation of TIS on cognitive functions remain limited. Existing research has validated the effectiveness of TIS principles [1][2][3], as well as its enhancement of cortical and motor functions [4][5][6][7][8], its impact on deep brain structures [6][8] and complex cognitive functions [9]. However, imaging results have shown considerable variabilities and discrepancies, such as enhanced cognitive performance is underpinned by increasing or decreasing neural activities in corresponding brain regions, or there is no direct relation between them at all. In this study we aim to shed light on these debates by investigating if theta-band TIS can enhance working memory performance and its underlying brain mechanisms.

As the "center of human cognition", working memory specifically functions as a system that simultaneously stores and processes information [10]. The change detection task [11][12], one of the most widely used paradigms in measuring working memory capacity, which comprises three stages: encoding, retention, and retrieval. During the encoding stage, participants encode and store a memory array containing one or multiple features, such as position, color, or shape. After a brief retention interval of a blank screen, participants respond to a detection array presented in the retrieval stage to determine whether it matches the memory array in specified features. Brain imaging outcomes have indicated the importance of the prefrontal cortex, primary visual cortex, and posterior parietal cortex in visual and spatial working memory, and that the right hemisphere contributes more to spatial working memory especially in conditions requiring the maintenance and manipulation of spatial information [13]. The effectiveness of tES in enhancing working memory has been well established. Theta band tACS over the frontal cortex can significantly improve working memory storage capacity [14][15], and overall performance [16][17]. Additionally, fMRI studies have demonstrated that synchronous tACS can enhance working memory performance by increasing brain activity and enhancing network connectivity strength [18]. Although behavioral studies have shown that TIS can enhance working memory performance [5], neuroimaging evidence supporting its impact on working memory processes and underlying neural mechanisms remains limited.

The mechanisms through which external electrical stimulation modulates neural activity are inherently complex, with existing research supporting multiple hypotheses. While tES is often regarded as an effective tool for modulating neural oscillations or enhancing cortical excitability [19][20][21], findings from neuroimaging studies remain inconsistent, particularly regarding its effects on higher-order cognitive functions. In research on the effects of tACS on working memory, both the increases and decreases of BOLD signals and functional connectivity strength have been testified [18][22][23]. Similar debates exist regarding the effects of TIS on cognitive functions. TIS may modulate network activity and enhance cognitive performance by increasing BOLD signals or functional connectivity strength [6][8][24]. Conversely, it may also reduce BOLD signals and functional connectivity strength while still achieving similar outcomes [9]. Based on these findings, we propose several hypotheses on the mechanisms by which TIS influences cognitive performance: (1) TIS may affect cognitive outcomes through direct modulation of BOLD signals at the stimulation target, without significantly altering functional connectivity; (2) TIS might exert its effects primarily by enhancing functional connectivity; or (3) due to the complex interplay between neural oscillations, BOLD signals, and functional connectivity, TIS could enhance cognitive performance through a nonlinear and multifactor mechanism.

Therefore, in this study we used the change detection task to assess spatial working memory performance, targeting the middle frontal gyrus and inferior parietal lobules of the right hemisphere as targets of TIS [5][14][15][18][25], combined with functional magnetic resonance imaging(fMRI), to explore the effect of TIS on spatial working memory performance and its neural mechanisms. In the neuroimaging outcomes, we predict that TIS will lead to a change in BOLD signals of stimulation targets and regions within the working memory network, accompanied by an enhancement in functional connectivity. We expect at least one of these neural changes correlate with improvements in behavioral performance.

Participants

A total of 31 (16 females) healthy participants were recruited in the age range of 18–26 years (mean age: 22.62 ± 2.06 years). All participants were right-handed, had normal visual acuity or corrected visual acuity, and were not majoring in psychology. They had no long-term exercise habits, no history of brain injury or craniotomy, no personal or family history of psychiatric or neurological diseases. Additionally, they had no metal or electronic device implants in the body, were not pregnant or taking medication during the experiment. None had participated in other transcranial stimulation experiments within three months prior to our study.

Experimental Design

Our experiment was a within-subject study with 2(stimulus type: TIS, sham stimulation) × 2(measurement time: pre-test, post-test) design. To wash out the effects of TIS, each participant was required to visit twice with a 7-d interval, one with TIS and the other with sham stimulation (Fig. 1a). Each visit scheduled at the same time of a day. During each visit, participants would receive two functional scanning sessions before 20min of offline TIS, immediately followed by another two sessions as post-test. The order of TIS stimulus and tasks was balanced among participants using the ABBA method.

Change-detection Tasks

We generated 3000 6×6 matrix images with 320×320 pixels in Matlab2019a (MathWorks, Natick, MA, United States) software, with black background cells and white stimulus cells. After removing images with adjacent, symmetrical, or shape-regulated white cells, the remaining images were used as encoding or detecting stimulation during the retention or retrieval stages. Task presentation and behavioral data recording were performed by E-Prime 3.0 (Psychology Software Tools, Sharpsburg, MD, United States). Participants were required to quickly and accurately judge whether the positions of the white cells in the retrieval stage were consistent with those in the encoding stage. In the low memory load trials, only one white cell was presented in both the retention and retrieval stage, while in the high memory load trials, there were 8 white cells in the retention stage and 4 white cells in the retrieval stage (Fig. 1b). A total of 272 trials were included in the two types of memory load, which were randomly divided into four sessions, each containing 68 trials, including 20 trials with low memory load (as a baseline) and 48 trials with high memory load. Randomized inter-trial interval (ITI) times (2-4.5 s, M = 2 s) were generated using Optseq 2 software to meet the requirements of fMRI scanning.

Temporal Interference Stimulation

TIS was performed using the non-invasive deep interference stimulation device (HD-IFS) developed by Soterix Medical, United States. We chose the middle frontal gyrus (MFG) and the inferior parietal lobule (IPL) in the right hemisphere as our targets, corresponded to F4 and P4 in the EEG 10–20 system (Fig. 1c) [5][18]. The electrodes were placed in parallel around the target area with a 5cm distance between the centers of each other. The frequency was set at 2000Hz and 2006Hz, the peak-to peak current was 2mA, stimulation waveform was B-sine. We set the frequency difference at 6Hz because many studies have previously found that theta tACS can regulate the WM process [14][15][18]. Each stimulation would last 20min with 15s fade-in and -out at the beginning and the end.

For sham stimulation, the electrodes were placed at the same site as TIS, but applied no current during the whole process. That was because TIS would bring no noticeable side effects such as pain, itching, fatigue or visual hallucinations etc. to subjects, so we opted to use a completely inactive sham stimulation to avoid the after-effect any short-term current may cause.

MRI Acquisition

The scanning of all participants was conducted using a 3.0 Tesla Siemens MAGNETOM Prisma whole-body MRI scanner, fitted with a 64-channel head coil (Siemens, Munich, Germany). The task-based images were collected using a gradient-echo EPI sequence (TR = 2100 ms; TE = 27 ms; FOV = 210×210 mm²; flip angle = 80°; voxel size = 3 × 3 × 3 mm³; 40 contiguous oblique axial slices). High-resolution structural images were collected using a 3D MPRAGE (magnetization-prepared rapid acquisition gradient echoes) pulse sequence (TR = 2300 ms; TE = 2:98 ms; flip angle = 9°; FOV = 256 × 256 mm²; voxel size = 1 × 1 × 1 mm³; 176 contiguous slices).

Behavioral data analysis

A two-way repeated-measures ANOVA (2×2) was used to investigate the main effects and their interaction for ACC and RT of stimulus type (TIS, sham), measurement time (pre-test, post-test). Effect sizes were measured by calculating the partial eta squared (η_p²) for two-way repeated-measures ANOVAs and Cohen’s d for post hoc paired-sample t-tests.

Image analysis

We performed preprocessing and statistical analysis on fMRI images in SPM12 (Institute of Cognitive Neurology, London, UK. http://www.fil.ion.ucl.ac.uk), under MATLAB (Mathworks Inc., Natick, MA, USA). We removed four lead-in scans in each session, performed slice timing correction, and then corrected head motion by realigned the remaining images to the middle slice. We used T1 images as an intermediate to coregister the mean functional images and normalize them to Montreal Neurological Institute (MNI) 152 standard space template. Spatial smoothing was applied using a Gaussian kernel filter with a full width at half maximum (FWHM) of 6mm. A high-pass filter with a cutoff frequency of 128 seconds was applied to remove low-frequency drift in the fMRI data.

Low memory condition was set as baseline, so in all subsequent analyses, results were reported by default as high memory load minus the baseline. We performed subject-level analysis using the general linear model (GLM), and random effects analysis at group level. We used a two-way repeated-measures ANOVA to observe the main effect of stimulus type and measurement time and their interaction during the three stages (encoding, retention, retrieval).

We defined 10mm radius spheres using center coordinates of TIS targeted regions as ROIs [18] and extracted their effect value using MarsBar [26], and correlated them with behavioral data to explore whether the changes of BOLD signals were associated with working memory performance. We then investigated if TIS modulates functional connectivity of working memory network using generalized psychophysiological interaction (gPPI) [27] in CONN version 22a [28].

Activation were threshold at p < 0.005, voxel-level uncorrected. Significant clusters were identified only when they survived a p < 0.05, FWE cluster-level corrected in whole brain analysis and region-of-interest (ROI) analysis. Coordinates of significant clusters peaks and subpeaks in each effect were listed in the Tables in standard MNI space, and was read automatically by AAL3 [29] in SPM12. In functional connectivity, cluster-level inferences were based on parametric statistics from Gaussian Random Field theory, and results were thresholded using a combination of a cluster-forming p < 0.005 voxel-level threshold, and a familywise corrected p-FDR < 0.05 cluster-size threshold.

Statistical analysis

All statistical analyses were conducted using R for Windows 4.2.3 via R Studio, plots were generated by the ggplot2 package. Image visualization was conducted using MRIcroGL [30] (https://www.nitrc.org/projects/mricrogl/). We conducted the Shapiro-Wilk normality test on all behavioral data and ROI effect value, and log-transformed all data didn’t conform to a normal distribution.

Behavioral Results

We performed two-way repeated-measures ANOVAs of 2 (stimulus type) × 2 (measurement time) for ACC and RT respectively. In all memory load conditions, we observed a significant main effect of measurement time (F_(1,60) = 26.79, p<0.001, η_p² = 0.47) in RTs, but no main effect of stimulus type (F_(1,60) = 0.95, p = 0.338, η_p² = 0.03) or interaction effect (F_(1,120) = 3.24, p = 0.082, η_p² = 0.09). When focusing on the high memory load condition, there were both a significant main effect of measurement time (F_(1,60) = 20.11, p<0.001, η_p² = 0.40) and an interaction effect (F_(1,120) = 4.48, p = 0.043, η_p² = 0.13), but the main effect of stimulus type remained non-significant (F_(1,60) = 1.45, p = 0.237, η_p² = 0.05). Only under TIS condition are the RTs in the post-test faster than those in the pre-test in high memory load condition (t₍₃₀₎=-4.48, p<0.001, d=-3.17 ), the RT difference of pre-test minus post-test was also bigger (t₍₃₀₎ = 2.12, p = 0.043, d = 0.38).(Fig. 1d)

For accuracy, there was no significant main effect of stimulus type (F_(1,60) = 0.44, p = 0.513, η_p² = 0.01), measurement time (F_(1,60) = 0.09, p = 0.768, η_p² = 0.003) or interaction (F_(1,120) = 0.73, p = 0.4, η_p² = 0.02) (Supplementary Table S1).

Task evoked activity

Brain activity evoked by different memory load tasks showed similar and expected patterns, either with or without TIS (Fig. 2 and Supplementary Table S2, Fig S1). In high memory load condition, BOLD signal increased mainly in visual cortex, sensory motor cortex, supplementary motor area, frontal cortex, parietal cortex, temporal cortex, fusiform, angular, insula, and cerebellum (Table 1). Default mode network (DMN) condition was significantly inhibited throughout all three stages in high memory load, especially in encoding and retrieval. However, we didn’t observe a similar pattern in low memory load condition in middle and superior frontal gyrus, as well as anterior cingulate cortex, indicating that a more demanding task would cause a greater deactivation in DMN (Supplementary Fig S2), which consisted with previous research on working memory [18][31].

Table 1
Brain Activation In RT Improved Paricipants
Region	Cluster size	Peak Z value	Peak MNI coordinates
Encoding Interaction
R:SMA,SFG,PCL,PreCG	88	3.94	9,-22,71
Encoding TIS:Pre-test > Post-test
SMA, R:SFG,PreCG,PCL	68	3.55	18,-10,68
Retention TIS:Pre-test > Post-test
L:IPL,ANG,SMG,SPG	154	4.39	-45,-40,32
R:INS,IFGOper,OLF,IFGtriang,MFG	62	3.86	39,26,-4
L:PUT,INS,CAU,IFGtriang,PAL,IFGOper	78	3.85	-21,17,-4
L:ITG,MTG,FFG,IOG	62	3.77	-51,-58,-16
ACG,SFGmed	82	3.33	-6,35,23

Whole-brain analysis in selected participants with shorter RT: Individual variability in TIS effects

We first performed two-way ANOVAs of stimulus type and measurement time (2×2) for the activation across the three stages of the change detection task (N = 31), and observed no significant main effect or interaction effect in the whole brain analysis. As a non-invasive external stimulation, the effect of TIS could be influenced by individual differences among participants, such phenomenon has already been improved in research using TMS, tDCS and tACS modulating neural activities or cognitive functions [32][33][34], and participants with lower baseline performance would benefit more from tACS stimulation [35]. Therefore, we conducted two-way repeated-measures ANOVAs on activation during three stages for participants who showed an improvement in RT (N = 19). During the encoding stage, a significant interaction was observed in the right SMA, SFG, PCL and PreCG, and the change in BOLD signals induced by TIS was significantly greater than any potential BOLD signal changes associated with sham stimulation (Fig. 3a, Table 1). Only under TIS condition, there was a significant decrease in BOLD signals during the encoding and retention stages in post-test, while no such decrease appeared between pre-test and post-test in sham stimulation condition. The change was observed in the SMA and sensory motor cortex during the encoding stage, and in the bilateral prefrontal cortex, left IPL, left SPL during the retention stage (Fig. 3a).

Although research on working memory and TIS is currently limited, existing studies on episodic memory and TIS have also reported a reduction in BOLD signal induced by TIS [9]. Our findings are consistent with previous human studies.

ROI analysis: Changes of BOLD signals correlated with behavioral performance

We extracted the effect sizes from the F4 and P4 targets across the three stages of working memory under different conditions. Paired-sample t-tests were performed on the effect sizes for each stage under each condition. Our analysis revealed that during the retention stage, TIS led to a significant decrease in BOLD signal at P4 in the post-test(t₍₃₀₎ = 2.04, p = 0.049, d = 0.36), while during the retrieval phase, there was a significant increase in BOLD signal at P4 in the post-test(t₍₃₀₎=-2.19, p = 0.037, d=-0.39) (Fig. 3b). To better explore whether these changes in BOLD signals at the stimulation targets were associated with behavioral changes, we conducted paired-sample correlation analyses between BOLD signal changes and task performance changes. We found a significant negative correlation between changes in BOLD signal at F4 region before and after TIS and RT across all participants (N = 31) (r₍₃₁₎=-0.48, p = 0.006) during encoding stage (Fig. 4a). No correlation was observed at P4 region (r₍₃₁₎=-0.27, p = 0.14). Among the participants with shorter RT (N = 19), one cluster significantly deactivated in TIS post-test (which including the left IFG and MFG, FFG, and IOG), showed a positive correlation between its BOLD signal change and accuracy change during retention stage (r₍₁₉₎ = 0.51, p = 0.02)(Fig. 4a). No correlation was observed in neither F4 region (r₍₁₉₎=-0.09, p = 0.68) or P4 region (r₍₁₉₎=-0.03, p = 0.89).

TIS Modulated Functional Connectivity of Working Memory Network

We then investigated if TIS modulates functional connectivity of working memory network using generalized psychophysiological interaction(gPPI) [27]. We performed seed-based analysis by taking IPL and MFG regions as seeds separately, to test whether they influenced network activity as targets of TIS. During the retention stage, stronger functional connectivity was observed between IPL region and the right angular gyrus, SPL, LOC and SMG in the post-test of TIS. A similar pattern was also found in the comparison between TIS and sham stimulation, where TIS induced stronger functional connectivity between P4 and these regions (Fig. 4b, Table 2). No significant changes in functional connectivity were observed during other stages or conditions. No correlation was observed between functional connectivity value and behavioral data.

Table 2
Functional Connectivity During Retention Stage
Region	Cluster size	Peak MNI coordinates
Seed: IPL TIS(Pre-Post) > Sham(Pre-Post)
R:SPL,ANG,LOC,SMG	150	42,-48,44
TIS:Pre-test > Post-test
R:ANG,SPL,LOC,SMG	150	36,-58,54

Our results demonstrate that TIS can modulate the BOLD signal and the working memory network through targeted regions (right MFG and IPL), and improve performance in high memory load visual\spatial working memory task. The effects of TIS may vary among individuals and could exert distinct aftereffects on different stages of working memory. Our study first revealed that participants who showed shorter RT exhibited decreased BOLD signals during the encoding and retention stages. This reduction was observed in the SMA, PreCG and PoCG during encoding, while during retention, it was evident in the bilateral prefrontal cortex, left IPL, and SPL. Although no significant changes were identified in whole-brain analyses at the stimulation targets F4 and P4, a notable correlation was found between BOLD signal decrease induced by TIS at F4 and improvements in behavioral performance. This finding confirms that TIS influenced behavioral outcomes via its effects on the stimulation target regions. Furthermore, functional connectivity analyses using P4 as a seed point revealed that TIS significantly enhanced the connectivity between P4 and SPL, SMG, and LOC during the retention stage. These results demonstrate that TIS can modulate the activity of the working memory network by influencing its stimulation targets. Our findings provide a reference framework for future mechanistic studies of TIS.

In the neuroimaging results, participants with shorter RTs exhibited significant deactivation at whole-brain level during the encoding and retrieval stages. This aligns, to some extent, with previous studies suggesting that TIS leads to decreased BOLD signals [1][9]. The decrease of BOLD signals across different brain regions during various stages of working memory may be related to the differential involvement of brain regions that play a predominant role at each stage. Previous research has highlighted the IPL's critical role in the retention phase of working memory, sustained attention, and spatial working memory [36][37][38]. Consistently, our findings demonstrated a significant reduction in the BOLD signal of the left IPL at the whole-brain level during the retention stage, as well as a significant decrease in the BOLD signal of the right IPL in the ROI analysis. These changes align with the established functions of the IPL. Previous studies have shown that frontal beta and theta oscillations can exhibit a negative correlation with BOLD signal [39][40]. Similarly, we propose that TIS may enhance network efficiency by increasing theta synchrony between the frontal and parietal regions[42], especially in IPL itself, which is reflected as BOLD signal reduction.

Although no direct correlation was found between neural activity in P4 and behavioral performance during the retention stage, we identified a negative correlation between RT and the activity of F4, and a positive correlation between response accuracy and activity in the IFG,MFG, and IOG — regions that are also part of the working memory network. This suggests that TIS may not simply modulate the BOLD signal at the targeted regions but may also influence the larger working memory network. Therefore, our study does not support the view that TIS influences brain activity by enhancing BOLD signals at the target or surrounding brain regions, or by recruiting additional neural resources. Instead, it suggests that TIS reduces the potential metabolic demands of the target and core brain regions, thereby enhancing the processing efficiency of the working memory network. This perspective has already been demonstrated in previous studies[9]. Our conclusions further support that this reduction in demand is associated with behavioral performance. This suggests that the effects of TIS may be a network phenomenon, a concept that has already been substantiated by research at the cellular level.[43].

Additionally, the deactivation of SMA and sensory cortex during the encoding stage might be attributed to the aftereffects of asynchronous TIS, which may induce task-driven spontaneous regulation [44], enhancing the efficiency of other regions by reducing BOLD activity in these areas. However, we did not find direct evidence linking these changes to behavioral performance. Previous TMS and tES studies have demonstrated that factors like state dependency and inter-individual variability can influence the effects of stimulation [44][45][46]. We propose that TIS might also be subject to such influences.

During the retention stage, P4 exhibited significantly enhanced functional connectivity with the SPL, LOC, and SMG, further supporting the idea that TIS can modulate larger brain networks beyond the stimulation targets. This network-level modulation may contribute to the observed improvements in behavioral performance. We speculate that TIS might suppress activity in regions less critical to the task — such as the SMA, which was inhibited during encoding—to enhance the efficiency of network activity. This result further reinforces our viewpoint and previous research[43], suggesting that the effects of TIS may enhance processing efficiency by reducing the resource consumption of certain brain regions and influencing the connectivity of the entire brain network.

However, no direct correlations were found between functional connectivity and behavioral performance. Instead, significant correlations were observed between RT and WM-related brain regions that were strongly activated during the retention stage. Additionally, although no significant BOLD signal changes were observed at F4 and P4 at the whole-brain level, a notable decrease in P4’s effect size during the retention stage was identified. This not only aligns with previous tACS and TIS research on working memory [1][9][18], but also suggests that the effects of TIS may not result from a simple linear modulation of neural activity at the target regions or changes in functional connectivity strength. Instead, the underlying mechanisms may be more complex, requiring further exploration using multimodal approaches.

The effect sizes at the stimulation target and the strength of functional connectivity varied across different stages of the task, with whole-brain analyses revealing similar patterns. This may be due to the distinct frequency bands and hemodynamic signal patterns associated with different stages of working memory [16][47][48]. For instance, research has shown that frontal theta oscillations negatively correlate with BOLD signals, whereas alpha and beta bands may exhibit positive correlations[39]. The differential contributions of various frequency bands to working memory functions, particularly the role of gamma oscillations during the maintenance phase has also been highlighted [49]. We speculate that the increased effect size observed in the P4 region during the retrieval phase may be attributed to the fact that theta oscillations are not the sole activity governing this stage. Previous research has demonstrated that other oscillatory bands are also engaged during retrieval [50][51]. Theta band stimulation might compel other oscillatory activities to recruit additional neural resources, thereby increasing the BOLD signal. Future studies should consider applying stage-specific stimulation tailored to different stages of working memory to gain a deeper understanding of the interaction between working memory processes and TIS mechanisms.

As with other neuromodulation techniques, the effects of temporal interference stimulation (TIS) are likely to exhibit interindividual variability. Previous research on tACS, tDCS, and TMS have identified various factors contributing to this variability, including endogenous oscillations, skull thickness, individual neural architecture, and circadian rhythms, which have led to some participants showing no physiological or behavioral changes, often referred to as "non-responsiveness" [44][45][46][52]. In our study, approximately 61.29% (19 out of 31) of participants demonstrated significant behavioral improvements and notable BOLD signal changes at the whole-brain level compared to the full sample. This proportion aligns with previous findings in non-invasive brain stimulation research [46][52], supporting the notion that individual differences play a critical role in the efficacy of these interventions.

Our study has several limitations. First, we did not employ TIS synchronized with fMRI, which prevents us from directly observing TI's effects on BOLD signals. The mechanisms by which TIS influences neural activity remain unclear, and the duration and consistency of its aftereffects are still debated. While asynchronous stimulation can provide stronger evidence of TIS's efficacy in enhancing cognitive functions, it may also introduce more uncontrolled variables. This could be one of the reasons for the differing patterns of BOLD signal changes across various stages. Second, the high memory load in our task design may have been overly challenging, potentially explaining the lack of significant improvement in accuracy. Future studies could consider using tasks with more appropriate difficulty levels to better investigate the mechanisms of TIS. Additionally, working memory is inherently sensitive to external stimulation. We applied theta-band TIS across all three stages of the change detection task but did not account for the distinct oscillatory of each stage, so we were unable to determine whether different stages exhibit different endogenous responses to TIS. Finally, we cannot ascertain whether theta-band TIS induces responses similar to those of tACS. The complex relationships between BOLD signals, neural oscillations, and functional connectivity remain unverified. Although we speculate that these relationships are non-linear, we lack the evidence to demonstrate how TIS affects these three aspects. Future research may need to incorporate simultaneous EEG and fMRI, as well as synchronized TIS and fMRI, to better explore these questions.

In summary, our study demonstrates the effecacy of TIS in enhancing spatial working memory performance and provides evidence of its underlying mechanism from a neuroimaging perspective. We propose that TIS does not enhance neural activity by increasing the BOLD signals at the target and associated brain regions. Instead, it reduces the resource consumption of these regions, lowering BOLD signals, which in turn leads to enhanced functional connectivity and improved efficiency across the entire brain network. This mechanism may be suggested by previous cellular-level studies. The correlation between the reduction in BOLD signals and behavioral performance further supports this conclusion. Our findings are also consistent with previous research on other neuromodulation techniques, which show that they alter neural circuit activity and improve cognitive performance. Future studies on the effects of TIS on cognitive functions and its impact on deep brain regions may find these findings valuable as a reference.

Ethic approval and consent to participate

This experiment was approved by the Ethics Committee of Shanghai University of Sport. All participants were informed about the experimental processes, durations, precautions, potential risks, and their right to withdraw freely at any stage of the experiment. They also acknowledged the ethical guideline of the study and provided written consent by signing the experimental checklist and informed consent form.

Consent for publication

The authors declare the consent for publication.

Availability of data and materials

Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. This study did not generate any new unique datasets or code.

Competing interests

The authors declare they have no conflict of interest related to this paper.

Funding

This research was funded by the National Natural Science Foundation of China (11932013, 62472319) and the Humanities and Social Sciences Fund of the Ministry of Education (21YJA190013).

Authors’ contributions

Y.F.Z., K.H., J.J.L., J.L., J.Z., and Y.L. conceived and designed the experiments. Y.S.C., K.H., Y.F.Z., and M.Z. performed the experiments. Y.S.C., J.Y.Z., L.Y.L., J.H.S., H.G., and M.L. provided methodological support and analyzed the data. Z.Y.Q. and J.Q.L. provided technical support. Y.S.C., K.H., J.Z., and Y.L. wrote the manuscript.

Acknowledgements

Not applicable.

Grossman N, Bono D, Dedic N, Kodandaramaiah SB, Rudenko A, Suk H-J, et al. Noninvasive Deep Brain Stimulation via Temporally Interfering Electric Fields. Cell. 2017;169(6):1029–e104116. 10.1016/j.cell.2017.05.024.
Zhu X, Li Y, Zheng L, Shao B, Liu X, Li C, et al. Multi-Point Temporal Interference Stimulation by Using Each Electrode to Carry Different Frequency Currents. IEEE Access. 2019;7:168839–48. 10.1109/ACCESS.2019.2947857.
Song X, Zhao X, Li X, Liu S, Ming D. Multi-channel transcranial temporally interfering stimulation (tTIS): Application to living mice brain. J Neural Eng. 2021;18(3):036003. 10.1088/1741-2552/abd2c9.
Ma R, Xia X, Zhang W, Lu Z, Zhang X. High Gamma and Beta Temporal Interference Stimulation in the Human Motor Cortex Improves Motor Functions. Int J Psychophysiol. 2021;168. 10.1101/2021.03.26.437107.
Zhang Y, Zhou Z, Zhou J, Qian Z, Lü J, Li L, et al. Temporal interference stimulation targeting right frontoparietal areas enhances working memory in healthy individuals. Front Hum Neurosci. 2022;16:918470. 10.3389/fnhum.2022.918470.
Wessel MJ, Beanato E, Popa T, Windel F, Vassiliadis P, Menoud P, et al. Noninvasive theta-burst stimulation of the human striatum enhances striatal activity and motor skill learning. Nat Neurosci. 2023;26(11):2005–16. 10.1038/s41593-023-01457-7.
Qi S, Liu X, Yu J, Liang Z, Liu Y, Wang X. Temporally interfering electric fields brain stimulation in primary motor cortex of mice promotes motor skill through enhancing neuroplasticity. Brain Stimul. 2024;17(2):245–57. 10.1016/j.brs.2024.02.014.
Vassiliadis P, Beanato E, Popa T, Windel F, Morishita T, Neufeld E, et al. Non-invasive stimulation of the human striatum disrupts reinforcement learning of motor skills. Nat Hum Behav. 2024;8(8):1581–98. 10.1038/gt9fj4.
Violante IR, Alania K, Cassarà AM, Neufeld E, Acerbo E, Carron R, et al. Non-invasive temporal interference electrical stimulation of the human hippocampus. Nat Neurosci. 2023. 10.1038/s41593-023-01456-8.
Baddeley AD. Working memory: Theories, models, and controversies. Annu Rev Psychol. 2012;63:1–29. 10.1146/annurev-psych-120710-100422.
Phillips WA. On the distinction between sensory storage and short-term visual memory. 1974.
Luck SJ, Vogel EK. The capacity of visual working memory for features and conjunctions. Nature. 1997;390(6657):279–81. 10.1038/36846.
van Asselen M, Kessels RP, Neggers SF, Kappelle LJ, Frijns CJ, Postma A. Brain areas involved in spatial working memory. Neuropsychologia. 2006;44(7):1185–94. 10.1016/j.neuropsychologia.2005.10.005.
Jaušovec N, Jaušovec K. Increasing working memory capacity with theta transcranial alternating current stimulation (tACS). Biol Psychol. 2014;96:42–7. 10.1016/j.biopsycho.2013.11.006.
Jaušovec N, Jaušovec K, Pahor A. The influence of theta transcranial alternating current stimulation (tACS) on working memory storage and processing functions. Acta Psychol (Amst). 2014;146:1–6. 10.1016/j.actpsy.2013.11.011.
Alekseichuk I, Turi Z, Amador de Lara G, Antal A, Paulus W. Spatial working memory in humans depends on theta and high gamma synchronization in the prefrontal cortex. Curr Biol. 2016;26(12):1513–21. 10.1016/j.cub.2016.04.035.
Röhner F, Breitling C, Rufener KS, Heinze H-J, Hinrichs H, Krauel K, Sweeney-Reed CM. Modulation of working memory using transcranial electrical stimulation: A direct comparison between tACS and tDCS. Front Neurosci. 2018;12:761. 10.3389/fnins.2018.00761.
Violante IR, Li LM, Carmichael DW, Lorenz R, Leech R, Hampshire A, Rothwell JC, Sharp DJ. Externally induced frontoparietal synchronization modulates network dynamics and enhances working memory performance. eLife. 2017;6:e22001. 10.7554/eLife.22001.
Antal A, Herrmann CS. Transcranial alternating current and random noise stimulation: Possible mechanisms. Neural Plast. 2016;2016:3616807. 10.1155/2016/3616807.
Bikson M, Esmaeilpour Z, Adair D, Kronberg G, Tyler WJ, Antal A, Datta A, Sabel BA, Nitsche MA, Loo C, Edwards D, Ekhtiari H, Knotkova H, Woods AJ, Hampstead BM, Badran BW, Peterchev AV. Transcranial electrical stimulation nomenclature. Brain Stimul. 2019;12(6):1349–66. 10.1016/j.brs.2019.07.010.
Wu L, Liu T, Wang J. Improving the effect of transcranial alternating current stimulation (tACS): A systematic review. Front Hum Neurosci. 2021;15:652393. 10.3389/fnhum.2021.652393.
Kar K, Ito T, Cole M, Krekelberg B. Transcranial alternating current stimulation attenuates BOLD adaptation and increases functional connectivity. Neuroscience. 2019. 10.1101/630368.
Abellaneda-Pérez K, Vaqué-Alcázar L, Perellón-Alfonso R, Bargalló N, Kuo MF, Pascual-Leone A, Nitsche MA, Bartrés-Faz D. Differential tDCS and tACS effects on working memory-related neural activity and resting-state connectivity. Front Neurosci. 2020;13:1440. 10.3389/fnins.2019.01440.
Zhu Z, Xiong Y, Chen Y, Jiang Y, Qian Z, Lu J, Liu Y, Zhuang J. Temporal interference (TI) stimulation boosts functional connectivity in human motor cortex: A comparison study with transcranial direct current stimulation (tDCS). Neural Plast. 2022;2022:7605046. 10.1155/2022/7605046.
Salazar RF, Dotson NM, Bressler SL, Gray CM. Content-specific frontoparietal synchronization during visual working memory. Science. 2012;338(6110):1097–100. 10.1126/science.1224000.
Brett M, Anton JL, Valabregue R, Poline JB. Region of interest analysis using an SPM toolbox. NeuroImage. 2002;16(2):1140. 10.1016/S1053-8119(02)90338-2.
McLaren DG, Ries ML, Xu G, Johnson SC. A generalized form of context-dependent psychophysiological interactions (gPPI): A comparison to standard approaches. NeuroImage. 2012;61(4):1277–86. 10.1016/j.neuroimage.2012.03.084.
Nieto-Castanon A, Whitfield-Gabrieli S. CONN functional connectivity toolbox: RRID SCR_009550, release 22. Hilbert; 2022. 10.56441/hilbertpress.2246.5840.
Rolland B, Herve PY, Mangin JF, Sakka L, Velut S, Mazoyer B, Tzourio-Mazoyer N. A multi-atlas based method for automated anatomical labeling of sulco-gyral anatomy in the MNI space. NeuroImage. 2020;206:116321. 10.1016/j.neuroimage.2019.116321.
Rorden C, Brett M. Stereotaxic display of brain lesions. Behav Neurol. 2000;12(4):191–200. 10.1155/2000/421719.
Cai W, Ryali S, Pasumarthy R, Talasila V, Menon V. Dynamic causal brain circuits during working memory and their functional controllability. Nat Commun. 2021;12(1):3314. 10.1038/s41467-021-23509-x.
Labruna L, Stark-Inbar A, Breska A, Dabit M, Vanderschelden B, Nitsche MA, Ivry RB. Individual differences in TMS sensitivity influence the efficacy of tDCS in facilitating sensorimotor adaptation. Brain Stimul. 2019;12(4):992–1000. 10.1016/j.brs.2019.03.008.
Gießing C, Alavash M, Herrmann CS, et al. Individual differences in local functional brain connectivity affect TMS effects on behavior. Sci Rep. 2020;10:10422. 10.1038/s41598-020-67162-8.
Elyamany O, Leicht G, Herrmann CS, Mulert C. Transcranial alternating current stimulation (tACS): From basic mechanisms towards first applications in psychiatry. Eur Arch Psychiatry Clin Neurosci. 2021;271(1):135–56. 10.1007/s00406-020-01209-9.
De Paolis ML, Paoletti I, Zaccone C, et al. Transcranial alternating current stimulation (tACS) at gamma frequency: an up-and-coming tool to modify the progression of Alzheimer’s Disease. Transl Neurodegener. 2024;13:33. 10.1186/s40035-024-00423-y.
Husain M, Nachev P. Space and the parietal cortex. Trends Cogn Sci. 2007;11(1):30–6. 10.1016/j.tics.2006.10.011.
Alain C, He Y, Grady C. The contribution of the inferior parietal lobe to auditory spatial working memory. J Cogn Neurosci. 2008;20(2):285–95. 10.1162/jocn.2008.20014.
Hu H, Li A, Zhang L, Liu C, Shi L, et al. Goal-directed attention transforms both working and long-term memory representations in the human parietal cortex. PLoS Biol. 2024;22(7):e3002721. 10.1371/journal.pbio.3002721.
Michels L, Bucher K, Lüchinger R, Klaver P, Martin E, Jeanmonod D, Brandeis D. Simultaneous EEG-fMRI during a working memory task: Modulations in low and high frequency bands. PLoS ONE. 2010;5(4):e10298. 10.1371/journal.pone.0010298.
Hanslmayr S, Volberg G, Wimber M, Raabe M, Greenlee MW, Bäuml KHT. The relationship between brain oscillations and BOLD signal during memory formation: A combined EEG–fMRI study. J Neurosci. 2011;31(44):15674–80. 10.1523/JNEUROSCI.3140-11.2011.
Polanía R, Nitsche MA, Korman C, Batsikadze G, Paulus W. The importance of timing in segregated theta phase-coupling for cognitive performance. Curr biology: CB. 2012;22(14):1314–8. https://doi.org/10.1016/j.cub.2012.05.021.
Fries P. A mechanism for cognitive dynamics: neuronal communication through neuronal coherence. Trends Cogn Sci. 2005;9(10):474–80. 10.1016/j.tics.2005.08.011.
Caldas-Martinez S, Goswami C, Forssell M, Cao J, Barth AL, Grover P. Cell-specific effects of temporal interference stimulation on cortical function. Commun Biology. 2024;7(1):1076. https://doi.org/10.1038/s42003-024-06728-y.
Cattaneo Z, Rota F, Vecchi T, Silvanto J. Using state-dependency of transcranial magnetic stimulation (TMS) to investigate letter selectivity in the left posterior parietal cortex: a comparison of TMS-priming and TMS-adaptation paradigms. Eur J Neurosci. 2008;28(9):1924–9. 10.1111/j.1460-9568.2008.06466.x.
Neuling T, Rach S, Herrmann CS. Orchestrating neuronal networks: Sustained after-effects of transcranial alternating current stimulation depend upon brain states. Front Hum Neurosci. 2013;7:161. 10.3389/fnhum.2013.00161.
López-Alonso V, Cheeran B, Río-Rodríguez D, Fernández-del-Olmo M. Inter-individual variability in response to non-invasive brain stimulation paradigms. Brain Stimul. 2014;7(3):372–80. 10.1016/j.brs.2014.02.004.
Friese U, Köster M, Hassler U, Martens U, Trujillo-Barreto N, Gruber T. Successful memory encoding is associated with increased cross-frequency coupling between frontal theta and posterior gamma oscillations in human scalp-recorded EEG. NeuroImage. 2013;66:642–7. 10.1016/j.neuroimage.2012.11.002.
Stout JJ, George AE, Kim S, Hallock HL, Griffin AL. Using synchronized brain rhythms to bias memory-guided decisions. eLife. 2024;12:RP92033. 10.7554/eLife.92033.
Roux F, Uhlhaas PJ. Working memory and neural oscillations: Alpha–gamma versus theta–gamma codes for distinct WM information? Trends Cogn Sci. 2014;18(1):16–25. 10.1016/j.tics.2013.10.010.
Osipova D, Takashima A, Oostenveld R, Fernández G, Maris E, Jensen O. Theta and gamma oscillations predict encoding and retrieval of declarative memory. J Neurosci. 2006;26(28):7523–31. 10.1523/JNEUROSCI.1948-06.2006.
Jokisch D, Jensen O. Modulation of gamma and alpha activity during a working memory task engaging the dorsal or ventral stream. J Neurosci. 2007;27(12):3244–51. 10.1523/JNEUROSCI.5399-06.2007.
Vergallito A, Feroldi S, Pisoni A, Romero Lauro LJ. Inter-individual variability in tDCS effects: A narrative review on the contribution of stable, variable, and contextual factors. Brain Sci. 2022;12(5):522. 10.3390/brainsci12050522.

No competing interests reported.

Supplementary.docx

Download PDF

Editorial decision: Revision requested
05 Jan, 2026
Reviews received at journal
29 Dec, 2025
Reviews received at journal
17 Dec, 2025
Reviewers agreed at journal
11 Dec, 2025
Reviewers agreed at journal
10 Dec, 2025
Reviewers invited by journal
10 Dec, 2025
Editor assigned by journal
24 Nov, 2025
Submission checks completed at journal
24 Nov, 2025
First submitted to journal
23 Nov, 2025

You are reading this latest preprint version

Exploring the Mechanisms of Temporal Interference Stimulation in Enhancing Spatial Working Memory Performance through Frontoparietal Modulation

Status:

Version 1

Abstract

Figures

Background

Materials and Methods

Participants

Experimental Design

Change-detection Tasks

Temporal Interference Stimulation

MRI Acquisition

Behavioral data analysis

Image analysis

Statistical analysis

Results

Behavioral Results

Task evoked activity

Whole-brain analysis in selected participants with shorter RT: Individual variability in TIS effects

ROI analysis: Changes of BOLD signals correlated with behavioral performance

TIS Modulated Functional Connectivity of Working Memory Network

Discussion

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1