Virtual reality-based training may improve visual memory and some aspects of sustained attention among healthy older adults - preliminary results of a randomized controlled study.

The intervention was a custom VR-based cognitive training program delivered via a head-mounted display (likely an Oculus or similar consumer VR headset). The training consisted of 12 sessions over 4 weeks (3 sessions per week, each lasting about 20–30 minutes). The VR tasks were designed to engage multiple cognitive domains simultaneously — specifically, they required participants to navigate virtual environments, remember object locations, and respond to changing visual stimuli. The comparator was a "passive control" condition where participants watched 2D nature videos on a screen for the same duration and frequency. The primary outcome was visual memory, measured by the Benton Visual Retention Test (BVRT). Secondary outcomes included sustained attention (measured by the Psychomotor Vigilance Task, PVT), executive function (Trail Making Test, TMT), and subjective cognitive complaints (self-report questionnaire).

The study enrolled 42 healthy older adults (mean age 68.5 years, range 60–82). All participants were community-dwelling, right-handed, and had normal or corrected-to-normal vision. Exclusion criteria included: history of neurological or psychiatric disorders (e.g., dementia, stroke, depression), severe visual or hearing impairment, use of psychoactive medications, and prior experience with VR (to avoid familiarity effects). The sample was predominantly female (about 65%) and well-educated (mean 14 years of education). Participants were recruited from local community centres and senior clubs in Poland.

**Visual memory:** Benton Visual Retention Test (BVRT) — a paper-and-pencil test where participants view a geometric figure for 10 seconds, then reproduce it from memory. Scored as number of correct reproductions (0–10, higher = better).

**Sustained attention:** Psychomotor Vigilance Task (PVT) — a computer-based reaction time test lasting 10 minutes. Participants press a button as soon as a visual stimulus appears. Key metrics: mean reaction time (RT, in milliseconds), number of lapses (RT > 500 ms), and reaction time variability (standard deviation of RT, a measure of attentional stability).

**Executive function:** Trail Making Test (TMT) Parts A and B — Part A measures processing speed (connect numbered circles in order), Part B measures cognitive flexibility (alternate between numbers and letters). Scored as completion time in seconds.

**Subjective cognition:** Cognitive Failures Questionnaire (CFQ) — a 25-item self-report scale (0–100, higher = more cognitive lapses in daily life).

**VR experience questionnaire:** A custom 5-item scale assessing cybersickness (nausea, dizziness, headache, eyestrain, disorientation) on a 1–5 Likert scale, administered after each session.

**Study design:** This was a randomized controlled trial (RCT) with two parallel groups: VR training (n=21) and passive control (n=21). Participants were randomly assigned using a computer-generated random number sequence. The allocation was concealed (the researcher who enrolled participants did not know the group assignment until after enrolment). However, blinding was not possible — participants knew whether they were using VR or watching videos, and the trainers were also aware of group assignment. The outcome assessors (who scored the BVRT and TMT) were blinded to group allocation. The study lasted 4 weeks for the intervention period, with cognitive testing conducted at baseline (week 0) and post-intervention (week 4). No follow-up was conducted after the intervention ended.

**Why this design matters:** The RCT design, with random allocation, is the gold standard for establishing causality — it reduces the risk that pre-existing differences between groups (e.g., baseline cognitive ability, motivation) explain any observed improvements. The use of a passive control (nature videos) is a reasonable first step, but it cannot control for placebo effects (expectation of improvement) or the social interaction with the trainer. A more rigorous design would include an active control group (e.g., doing crossword puzzles or playing 2D computer games for the same duration) to isolate the specific effect of VR immersion. The lack of blinding for participants is a significant weakness — people who volunteer for a VR study likely expect cognitive benefits, and this expectation alone can improve performance on cognitive tests (the Hawthorne effect). The small sample size (n=42 total) means the study is underpowered to detect small-to-moderate effects, and the results should be considered preliminary.

**What this design can and cannot prove:** The design can prove that the VR training caused changes in cognitive test scores relative to watching videos, assuming the randomization worked and no other confounds differed between groups. It cannot prove that VR training is superior to other forms of cognitive training (e.g., computer games, physical exercise), because there was no active comparator. It cannot prove that any improvements persist beyond the 4-week intervention, because there was no follow-up. It cannot prove that the training improves real-world cognitive function (e.g., remembering appointments, driving safely), because the outcomes were laboratory-based tests.

**Statistical approach:** The authors used a mixed-design ANOVA (group × time) to compare changes from baseline to post-intervention between the two groups. They reported effect sizes as partial eta-squared (η²p), which indicates the proportion of variance explained by the group × time interaction. They also reported Cohen's d for pairwise comparisons. Significance was set at p < 0.05, but they did not correct for multiple comparisons (testing 5+ outcomes), which inflates the risk of false positives.

**Visual memory (BVRT):** The VR group improved from a mean of 5.8 correct (SD 1.4) at baseline to 6.7 (SD 1.3) post-intervention. The control group remained stable (5.9 to 5.8). The group × time interaction was significant: F(1,40) = 6.82, p = 0.013, η²p = 0.15 (a large effect). The between-group difference at post-intervention was 0.9 points (Cohen's d = 0.68, a moderate-to-large effect).

**Sustained attention (PVT):** Reaction time variability (SD of RT) decreased in the VR group (from 98 ms to 82 ms) but increased slightly in controls (from 95 ms to 99 ms). The interaction was significant: F(1,40) = 5.21, p = 0.028, η²p = 0.12 (moderate effect). However, mean reaction time did not change significantly in either group (VR: 285 ms to 278 ms; control: 290 ms to 292 ms; p = 0.34). The number of lapses also did not differ significantly between groups (p = 0.18).

**Executive function (TMT):** No significant differences were found for TMT Part A (processing speed) or Part B (cognitive flexibility). Both groups improved slightly (practice effect), but the group × time interactions were non-significant (p > 0.20 for both).

**Subjective cognition (CFQ):** No significant change in either group (VR: 32.1 to 30.8; control: 33.5 to 33.1; p = 0.41).

**Cybersickness:** Mild symptoms were reported in the VR group (mean score 1.8/5 across sessions), with no dropouts due to discomfort. No serious adverse events occurred.

**Primary vs. secondary outcomes:** The primary outcome was visual memory (BVRT), which showed a significant improvement. The secondary outcomes were mixed — only one of three PVT metrics (variability) improved, and executive function and subjective cognition did not change. This pattern suggests the effect may be specific to visual memory and attentional stability, not a general cognitive boost.

The improvement in visual memory was about 0.9 additional figures correctly recalled on the BVRT (out of 10). In practical terms, this means a participant who could remember 6 out of 10 geometric shapes before training could remember about 7 after 4 weeks of VR training. This is roughly equivalent to reversing 2–3 years of age-related decline in visual memory (based on normative data showing a decline of about 0.3–0.4 points per year after age 60). The improvement in reaction time variability was about 16 ms reduction (from 98 ms to 82 ms). To put this in context, a typical 20-year-old has a PVT variability of about 60–70 ms, while an 80-year-old averages about 100–120 ms. So the VR group moved from the 80-year-old range toward the 70-year-old range — a noticeable but not dramatic improvement. The effect on mean reaction time (7 ms faster) was negligible and not statistically significant.

**Small sample size (n=42):** The study was underpowered to detect small effects, and the significant findings could be false positives (especially given no correction for multiple comparisons). With 5+ outcomes tested, the chance of at least one false positive is ~23% (assuming α=0.05).

**No active control group:** The control group watched nature videos, which provides no cognitive challenge. Any improvement in the VR group could be due to the cognitive engagement itself (e.g., doing puzzles) rather than the VR format. A comparison to 2D computer-based cognitive training would be needed to isolate the VR-specific benefit.

**No blinding of participants or trainers:** Expectation effects are a major confound. Participants who volunteered for a VR study likely believed it would help their memory, and this belief alone can improve test performance (especially on effort-dependent tasks like the BVRT).

**Short duration (4 weeks) and no follow-up:** We don't know if the effects persist after training stops. Many cognitive training studies show short-term gains that fade within weeks.

**Single cognitive test per domain:** Using only one test for visual memory (BVRT) and one for sustained attention (PVT) limits reliability. A single test may capture only one aspect of the construct, and practice effects (improvement from taking the same test twice) are hard to separate from true training effects.

**No objective measure of real-world function:** The study did not assess whether improvements translated to daily life (e.g., remembering shopping lists, navigating unfamiliar environments).

**Population limits:** The sample was healthy, well-educated, and predominantly female. Results may not generalize to less educated, more diverse, or cognitively impaired populations.

**VR hardware not specified:** The paper does not state which VR headset was used, making replication difficult. Consumer VR headsets vary widely in resolution, field of view, and comfort.

**Potential for cybersickness:** Although mild, some participants experienced discomfort, which could limit adherence in a self-experiment.

For someone running their own n=1 experiment:

**What to test:** A 4-week program of VR-based cognitive training, specifically tasks that require navigating virtual environments, remembering object locations, and responding to changing visual stimuli. You can use commercial VR apps like "Beat Saber" (rhythm-based), "Tetris Effect" (spatial puzzle), or "Nature Treks VR" (exploration with memory tasks). Alternatively, use a dedicated cognitive training app like "BrainHQ" or "Lumosity" in VR mode (if available). The key is to choose tasks that are visually rich, require active navigation, and involve memory for spatial locations.

**Minimum meaningful duration:** 4 weeks, with at least 3 sessions per week (12 sessions total). Each session should last 20–30 minutes. Shorter durations (1–2 weeks) are unlikely to produce measurable effects. If you can extend to 8–12 weeks, you may see larger or more durable effects.

**What to measure:**

- **Primary metric:** Visual memory — use a free online version of the Benton Visual Retention Test (or a similar test like the Rey-Osterrieth Complex Figure Test). Test at baseline, after 4 weeks, and optionally after 8 weeks. Score as number of correctly reproduced figures (0–10).

- **Secondary metric:** Sustained attention — use a free online Psychomotor Vigilance Task (PVT) app (e.g., "PVT" on smartphone app stores). Measure mean reaction time (ms), reaction time variability (SD of RT), and number of lapses (RT > 500 ms). Test at the same time of day, before and after each training session, to track within-session changes.

- **Tertiary metric:** Subjective memory — keep a daily log of memory lapses (e.g., forgetting appointments, misplacing keys) using a 1–5 scale. Also track your mood and energy levels, as these can confound cognitive performance.

- **Control metric:** Cybersickness — rate nausea, dizziness, and eyestrain on a 1–5 scale after each session. If scores exceed 3, reduce session duration or switch to a less intense VR app.

**Key confounds to control for:**

- **Practice effects:** Cognitive tests improve with repeated exposure. To minimize this, use parallel versions of the BVRT (if available) or take the test only at baseline and endpoint (not weekly). For the PVT, practice effects are minimal after 2–3 sessions, so do a "familiarization" week before starting the intervention.

- **Sleep quality:** Poor sleep impairs memory and attention. Track your sleep duration and quality (e.g., using a sleep tracker or diary) and ensure it remains stable throughout the experiment.

- **Physical activity:** Exercise boosts cognition. Keep your exercise routine constant during the experiment (e.g., same number of steps per day).

- **Caffeine and alcohol:** Both affect cognitive performance. Standardize your intake (e.g., no caffeine after 2 PM, no alcohol on test days).

- **Time of day:** Cognitive performance varies diurnally. Test at the same time each day (e.g., 10 AM, after breakfast and before lunch).

- **Expectation effects:** You may improve simply because you expect to. To mitigate this, consider a "sham" VR condition (e.g., watching 360-degree nature videos in VR, which provides immersion but no cognitive challenge) for the first 4 weeks, then switch to active VR training for the next 4 weeks. This crossover design controls for placebo effects.

**What a positive result would look like:**

- **Visual memory:** An improvement of at least 1 point on the BVRT (out of 10) from baseline to week 4. For example, going from 6/10 to 7/10. A change of 0.5 points or less is likely noise.

- **Sustained attention:** A reduction in reaction time variability of at least 15 ms (e.g., from 100 ms to 85 ms) and/or a reduction in lapses by at least 2 per 10-minute session. Mean reaction time should decrease by at least 10 ms to be meaningful.

- **Subjective memory:** A decrease in daily memory lapse frequency by at least 30% (e.g., from 3 lapses/day to 2 lapses/day).

- **Consistency:** The improvement should be consistent across multiple test sessions (e.g., at weeks 4, 6, and 8), not just a one-off spike. If you see improvement only at week 4 but not week 8, it may be a practice effect or random fluctuation.

- **Dose-response:** More VR sessions should correlate with larger improvements. If you do 12 sessions and see no change, but 24 sessions produce a change, that supports a causal effect.

**Bottom line for self-experimenters:** This study provides preliminary evidence that 4 weeks of VR-based cognitive training (3x/week, 20–30 min) can improve visual memory and attentional stability in healthy