Neural changes in early visual processing after 6 months of mindfulness training in older adults

An emerging body of research suggests that mindfulness training enhances attentional performance by repeatedly engaging neural circuitry for information processing and attentional control. Mindfulness develops sustained attention via metacognitive monitoring and control to resist distraction, often using breath sensations as the initial focus, coupled with an equanimous orientation toward experience. This practice recruits attentional control processes and engages anterior cingulate and right dorsolateral prefrontal cortex networks. ERP measures allow tracing visual processing across sensory encoding (50–80 ms), early discrimination (100–200 ms), and later cognitive processing (200–400 ms). While short 8-week interventions have shown long-latency ERP changes (e.g., N2, P3), the stability of these effects and potential impacts on earlier stages (P1, N1) over longer training durations remain unclear. Aging is associated with slowed and diminished perceptual and attentional processing, including increased latency and reduced amplitude of P1/N1 (sensory/perceptual) and N2/P3 (later cognitive) components, likely due to deterioration of white matter in visual pathways and reduced top-down facilitation. Cognitive training can also modify long-latency ERPs and attentional performance in older adults. The current active-controlled longitudinal randomized controlled trial (RCT) in healthy older adults assessed immediate effects of an 8-week mindfulness training (T1–T2) and a 6-month follow-up (T3) to test whether effects shift earlier in the information processing stream with continued practice. Hypotheses were: both mindfulness training (MT) and computer-based cognitive training (CT) would improve attentional performance and elicit long-latency ERP changes after 8 weeks; if ERP changes arise from attention training per se, MT and CT should produce similar effects; and with 6 months of MT, ERP changes would shift earlier (e.g., P1/N1).

Prior mindfulness studies have reported changes in long-latency ERP components (N2, P3) indicative of faster deployment and enhanced allocation of attentional resources, suggesting modulation of top-down processes after short-term training. Cross-sectional comparisons of long-term practitioners versus controls also revealed alterations in short-latency P1 and N1 components alongside N2–P3, implying upstream influences on early sensory and perceptual stages with extensive practice. Cognitive models of mindfulness propose that as proficiency grows, attentional capacity can be allocated at increasingly earlier processing stages. Aging degrades perceptual and attentional processes: deterioration in white matter (e.g., magnocellular optic radiations) slows visual processing, reducing P1/N1 amplitudes and increasing latencies, while declines in inhibitory control elevate susceptibility to distraction and reduce sustained attention, reflected in altered N2/P3. Cognitive training in older adults improves attentional performance and increases N2/P3 amplitudes, but typically does not alter ERP latencies. Given overlap between attention training components in mindfulness and cognitive training, it is important to determine whether neurophysiological changes are due to shared attentional mechanisms or unique to mindfulness’ equanimity and non-elaborative attention features.

Design and participants: Active-controlled longitudinal randomized controlled trial in healthy adults over 60 years, blinded to condition labels (both described as attention training). Random allocation at 2:1 ratio to mindfulness training (MT) or computer cognitive training (CT) anticipating greater attrition in MT. Initial allocation: MT n=77; CT n=43. After attrition, n=81 completed all three timepoints (MT n=50; CT n=31). No remuneration; screened to exclude factors affecting cognition/EEG and to exclude prior mindfulness/meditation experience. Ethics approval obtained (University of the Sunshine Coast HREC A-15-748); informed consent per Declaration of Helsinki.

Interventions: Both programs had weekly group sessions plus daily home practice, starting at 20 min/day in week 1 increasing to 45 min/day by week 8. No required practice after the 8-week intervention; participants recorded practice time across 6 months; monthly check-ins reduced attrition.

• Mindfulness training (MT): Standardised mindfulness technique operationalising unbiased, non-elaborative, nonjudgmental attention to present-moment experience (initially breath sensations, using optional labeling such as “rising/falling” early in training, progressing to direct non-labeled awareness). Emphasizes attentional self-regulation and equanimity without ancillary components (e.g., yoga, relaxation, psychotherapy).
• Computer cognitive training (CT): Game-based tasks targeting similar attentional processes (sustained/selective attention, executive control, working memory). Included modified Eriksen flanker, visual search, task switching, divided attention, Corsi block, and card matching tasks. Participants practiced one task per session continuously (to train sustained attention); tasks changed weekly; accessible online any time. CT and MT matched for contact time and structural elements.

Assessments and tasks: Three timepoints: baseline (T1), post-8-week intervention (T2), and 6-month follow-up from training onset (T3). EEG recorded during a breath counting task and an AX Continuous Performance Task (AX-CPT). Anxiety and depression symptoms assessed at each timepoint; no participant had elevated symptoms. Estimated FSIQ derived at T1 via Wechsler Test of Adult Reading.

• Breath counting task (15 min): Participants counted in-out breath cycles from 1 to 21, reporting each via keypress; count of 21 indicated with a right-arrow; self-caught errors via left-arrow. A chest effort sensor captured respiration to validate performance. Breath counting accuracy calculated as percentage of correct cycles.
• AX-CPT: Serial presentation of angular letters (A, E, F, H, L, N, T, V, X, Y, Z) at random sizes (100–180 pt) on grey background for 200 ms, followed by fixation (500–1000 ms jitter). Respond quickly/accurately to X only when immediately preceded by A. 800 trials: 60 AX targets (7.5%), 60 A* (non-X following A), 60 *X (X not preceded by A). Practice block with feedback preceded experimental block. Behavioral measures: total errors (% of trials), reaction time (RT), and RT coefficient of variation (RTCV = SD/mean).

EEG acquisition and preprocessing: 128-channel BioSemi ActiveTwo, 1024 Hz sampling; vertical/horizontal EOG recorded. Data segmented −2000 ms to +4000 ms around cue A or target X for preprocessing padding; filtered 1–40 Hz (zero-phase Butterworth IIR); average reference; ICA used to remove ocular/cardiac artifacts after artifact trial rejection; residual artifacts removed on inspection. Less than 10% of trials rejected across timepoints. ERPs from correct trials, segmented −200 ms to +600 ms relative to stimulus onset for cue and target; baseline corrected using −200 to 0 ms.

Statistical analysis: Baseline between-group comparisons via independent t-tests (age, FSIQ, breath counting, AX-CPT) and chi-square (gender, handedness). Mixed-model ANOVAs: factors group (MT, CT) and time (T1, T2, T3) for breath counting and AX-CPT metrics. ERP effects tested using non-parametric two-tailed cluster-based permutation tests (5000 permutations), clusters defined as spatiotemporal patterns (min 2 significant channels) exceeding critical t; cluster mass computed as sum of t-values; significance at alpha=0.05 using max-cluster distribution; time window 0–600 ms. Effect sizes (Cohen’s d) computed over significant clusters. For interaction effects with 3-level time factor, pairwise interaction contrasts were formed (e.g., [MT T3–T1] – [CT T3–T1]). Area under the curve (AUC) across significant cluster channels/timepoints extracted and correlated with target RT (Pearson). Analyses conducted blinded to condition using Python (Pingouin 0.3.7); FDR controlled via Benjamini–Hochberg; only sub-critical p-values reported.

Engagement/training time: During 8-week intervention, CT trained longer per day than MT (CT mean 35.4±5.5 min/day vs MT 33.0±4.2; t(79)=2.21, p=0.030). At 6 months, 36.7% of CT and 58.8% of MT reported continued practice. Averaged over 6 months, CT trained ~3 min/day more than MT (CT 30.1±6.0 vs MT 27.2±5.4; t(69)=2.29, p=0.025).

Baseline: No significant pre-intervention group differences in age, gender, estimated FSIQ, handedness, breath counting accuracy, or AX-CPT performance.

Mindfulness (breath counting): Significant group × time interaction (F(2,154)=6.95, p=0.001, d=0.60). MT improved from T1 to T2 (t(49)=4.2, p=0.0001) with gains retained at T3 (T1–T3: t(49)=3.2, p=0.002). CT showed no significant change.

Attentional performance (AX-CPT): Significant time effects for errors (F(2,132)=11.64, p<0.0001, d=0.85) and RTCV (F(2,144)=11.79, p<0.0001, d=0.81); no significant change in RT (F(2,144)=2.87, p=0.06, d=0.40). No group × time interactions. Exploratory pairwise: MT reduced errors T1–T2 (t(43)=3.57, p=0.0004, g=0.65) and T1–T3 (t(43)=4.26, p=0.00006, g=0.82); CT showed no significant error changes.

ERP—Cue stimulus (A): Main effect of time, no main effect of group and no significant pairwise interaction tests. Post hoc within-group comparisons:

• MT: T1–T3 increased amplitude in early P1 latency (33–107 ms), left parieto-occipital (sum(t)=7.13×10^3, Pcluster=0.01, d=0.65); and increased N1 amplitude (155–190 ms), parietal-temporal (sum(t)=−5.6×10^3, Pcluster=0.045, d=0.54). T1–T2 N1 trend did not reach significance; no significant T2–T3 change.
• CT: T1–T3 increased amplitude spanning N1–P2 (178–235 ms), occipital to later centroparietal distribution (sum(t)=7.43×10^3, Pcluster=0.01, d=0.73). No significant T1–T2 or T2–T3 clusters.

ERP—Target stimulus (X): Main effect of time; no significant group main effect or pairwise interaction tests. Post hoc within-group:

• MT: T1–T3 increased P3 amplitude (230–400 ms), parietal at onset shifting to frontocentral by ~300 ms (sum(t)=1.05×10^4, Pcluster=0.01, d=0.36). Progressive increase from T1 to T2 to T3 observed.
• CT: No significant clusters for any time comparison.

Brain–behavior correlations (MT): Greater cue N1 AUC (155–190 ms) associated with shorter target RT (p=0.007, CI [0.06, 0.39]); greater target P3 AUC (230–400 ms) associated with shorter RT (reported as negative association; p=0.03, CI [−0.34, −0.02]); cue P1 AUC not associated with RT (p=0.44). In CT, cue N1–P2 AUC not associated with RT (p=0.38). Cue N1 AUC correlated with target P3 AUC in MT (r=−0.26, p=0.002), indicating linkage between enhanced perceptual discrimination and later post-perceptual processing. Overall, MT after 6 months showed enhancements at early (P1, N1) and late (P3) stages and these were behaviorally relevant; CT showed limited cue-related N1–P2 changes without behavioral linkage and no target-related effects.

The study tested whether mindfulness training in older adults enhances attentional processes and whether neural changes shift to earlier processing stages with continued practice. After 6 months of mindfulness practice, participants exhibited enhanced early sensory encoding (P1) and perceptual discrimination (N1) to a visual cue and increased late post-perceptual processing (P3) to a target, with correlations linking cue N1 to target P3 and both to faster responses. These findings indicate increased efficiency in bottom-up visual pathways and improved mobilization of top-down attentional resources during perceptual and post-perceptual stages. The early P1 effect likely reflects stimulus-driven inhibition of irrelevant networks to boost signal-to-noise in relevant networks, while N1 indexes enhanced discrimination facilitating subsequent processing. Aging-related slowing due to white matter degradation may be mitigated by mindfulness, potentially via improved white matter integrity, consistent with reports of increased connectivity in long-term practitioners. The pattern of effects aligns with cross-sectional findings in long-term meditators, suggesting that improvements in early attentional processing can manifest within 6 months rather than requiring years. The CT group, despite similar behavioral improvements in sustained attention and slightly greater practice time, showed only cue-related N1–P2 changes consistent with visual discrimination enhancements from game-based training, without target-related or behaviorally linked ERP changes. The absence of immediate (post-8-week) significant ERP changes but clear progressive waveform trends indicates that neural adaptations accrue over time, possibly clarified by the use of interventions stripped of ancillary components (e.g., relaxation, yoga) that can confound immediate effects in other programs.

Six months of mindfulness training in healthy older adults enhanced early sensory and perceptual processing of visual stimuli (P1, N1) and late post-perceptual processing (P3), accompanied by improved sustained attention. The coupling between enhanced cue-related N1 and target-related P3 and their association with faster responses suggests that early perceptual gains influence later cognitive processes, potentially supporting efficient stimulus–response integration. These results indicate that mindfulness may help counteract age-related declines in attentional processing and neural efficiency and that such effects can emerge within months of practice. Future research should include neuroimaging to directly assess white matter changes, incorporate no-training control groups to further rule out practice effects, examine dose–response relationships, and test generalization to clinical populations and other cognitive domains.

• No no-training passive control group; although ERPs are generally reliable and resistant to practice effects, their contribution cannot be fully excluded.
• Immediate post-intervention (8-week) ERP changes did not reach significance; effects emerged progressively by 6 months, limiting conclusions about short-term neural efficacy.
• CT participants engaged in slightly more daily training than MT; while unlikely to explain MT-specific ERP effects, differential engagement could influence outcomes.
• Cluster-based permutation interaction testing is challenging with multi-level factors; interaction inferences relied on pairwise contrast approaches.
• Sample limited to healthy older adults, which may affect generalizability to other age groups or clinical populations.