A randomized controlled trial of cognitive training using a visual speed of processing intervention in middle aged and older adults.

The researchers tested whether a specific computerized cognitive training program called "Road Tour" (commercially available from Posit Science) could improve or stabilize cognitive function in middle-aged and older adults. The intervention trains visual speed of processing — the ability to quickly identify and locate visual information in your peripheral vision while simultaneously paying attention to something in your central vision. Think of it as a video game where you must identify a target (like a car or truck) in the center of the screen while also locating a peripheral target (like a road sign), with the display time getting progressively shorter as you improve.

Participants were randomized into four groups:

**10-hour on-site training:** Five weekly two-hour sessions of the visual speed of processing game at the university lab

**14-hour on-site training:** The same 10-hour program plus four additional booster hours (two hours at week 11, two hours at week 35)

**10-hour at-home training:** The same program but done on the participant's own home computer, with phone support

**Attention control group:** Five weekly two-hour sessions of computerized crossword puzzles at the university lab (10 hours total)

The primary outcome was the Useful Field of View (UFOV) test, which measures how quickly you can process visual information across your entire field of vision. Secondary outcomes included several standard neuropsychological tests of processing speed, attention, executive function, and verbal fluency.

**Sample size:** 681 participants randomized; 620 completed the one-year follow-up (91% retention)

**Population:** Patients from general internal medicine and family medicine clinics at the University of Iowa

**Age:** Two age bands — 50–64 years old and 65 years or older

**Inclusion criteria:** Age 50+, at least two clinic visits in the past year, no ICD-9 codes indicating cognitive impairment (Alzheimer's, dementia, etc.), had a home PC with internet connection, no significant uncorrected vision issues, scored fewer than 3 errors on a mental status screening, lived within 30 miles of the project office

**Setting:** University of Iowa laboratory (on-site groups) or participants' homes (at-home group)

**Recruitment:** From 5,743 potential participants identified through electronic medical records; 681 enrolled after screening

The researchers used a battery of standard neuropsychological tests at baseline and again at one year:

**Useful Field of View (UFOV) test:** The primary outcome. Measures processing speed across the visual field. Participants identify a central target (car or truck) while simultaneously locating a peripheral target (road sign). Display time starts at 500ms and decreases. Lower scores (in milliseconds) = faster processing. This is the same test that the training game was designed to improve.

**Trail Making Test A (Trails A):** Measures visual scanning and processing speed. Participants draw lines connecting numbered circles in order (1-2-3...). Score is time to completion in seconds. Lower = faster.

**Trail Making Test B (Trails B):** Measures executive function and cognitive flexibility. Participants alternate between numbers and letters (1-A-2-B-3-C...). Score is time in seconds. Lower = faster.

**Symbol Digit Modalities Test (SDMT):** Measures processing speed and attention. Participants match symbols to numbers using a key. Score is number correct in 90 seconds. Higher = better.

**Stroop Color and Word Test:** Measures selective attention and processing speed. Three conditions: reading color names (Word), naming color patches (Color), and naming the ink color of incongruent color words (Color-Word, e.g., the word "RED" printed in blue ink). Score is number correct in 45 seconds per condition. Higher = better.

**Controlled Oral Word Association Test (COWAT):** Measures verbal fluency. Participants generate as many words as possible starting with a given letter (F, A, S) in 60 seconds each. Score is total words. Higher = better.

**Digit Vigilance Test (DVT):** Measures sustained attention. Participants cross out target digits (6 and 9) from rows of numbers. Score is time to completion. Lower = faster.

**Study design:** Four-arm, parallel-group randomized controlled trial (RCT).

**Randomization:** Participants were randomized within two age bands (50–64 and 65+) using a computerized algorithm with a 3:3:4:4 allocation ratio (favoring the two on-site training groups and the control group slightly over the at-home group). Random permuted blocks of sizes 4, 8, and 12 were used. Sequentially numbered opaque envelopes ensured allocation concealment.

**Blinding:** Baseline assessments were fully double-blinded — neither the participant nor the interviewer knew group assignment because envelopes remained sealed until after the initial visit. After randomization, participants knew their group (no sham training for controls), so this was a single-blind design for the intervention period and follow-up. Outcome assessors were not explicitly stated to be blinded at follow-up, which is a potential weakness.

**Duration:** Training occurred over 5–10 weeks (one two-hour session per week for on-site groups; at-home group completed sessions at their own pace). Follow-up assessment occurred at one year post-baseline.

**Statistical approach:** Linear mixed models were used with Blom rank transformations within age bands to normalize the skewed outcome data. This is appropriate for cognitive test data that often have ceiling and floor effects. Analyses were intention-to-treat (all 620 completers analyzed in their randomized groups). Effect sizes were reported as Cohen's d (standardized mean differences).

**What this design can prove:** The RCT design with an active attention control group (crossword puzzles) allows causal inference about the specific effects of visual speed of processing training beyond general cognitive stimulation or placebo effects. The inclusion of an at-home arm tests whether the intervention can be delivered remotely. The one-year follow-up assesses durability of effects.

**What this design cannot prove:** Because the control group received crossword puzzles (not a placebo with equivalent expectations), placebo effects cannot be fully ruled out — participants who knew they were getting the "real" training may have had different expectations than those doing crosswords. The study cannot determine whether effects persist beyond one year, whether training transfers to real-world outcomes (like driving safety or fall risk), or whether the effects are specific to visual speed of processing versus any cognitively engaging activity. The lack of a no-contact control group means we cannot assess natural age-related decline over the year.

**Major methodological weaknesses:**

No sham training for the control group (crossword puzzles are clearly different from the visual training)

Outcome assessors may not have been blinded at follow-up

The primary outcome (UFOV) is thematically very similar to the training task, raising the possibility of "training to the test"

9% attrition, though relatively low, could bias results if dropouts differed systematically

The sample was drawn from a single academic medical center in Iowa, limiting generalizability

**Primary outcome — UFOV (processing speed):**

All three training groups showed significant improvements compared to the control group (p < 0.05 for all)

Effect sizes (Cohen's d): -0.322 (10-hour on-site), -0.579 (14-hour on-site), -0.388 (10-hour at-home)

Negative d values indicate faster processing (lower UFOV scores)

Converted to years of protection against age-related decline: 3.0 years (10-hour on-site), 4.1 years (14-hour on-site), 3.4 years (at-home)

**Secondary outcomes — Processing speed and attention:**

*Trail Making Test A (processing speed):*

Significant improvements in all training groups (p < 0.05)

Effect sizes: d = -0.204 to -0.265

Years of protection: 2.2 to 3.5 years

*Trail Making Test B (executive function):*

Significant improvements in all training groups (p < 0.05)

Effect sizes: d = -0.225 to -0.320

Years of protection: 1.5 to 2.0 years

*Symbol Digit Modalities Test (processing speed):*

Significant improvements in all training groups (p < 0.05)

Effect sizes: d = 0.263 to 0.351

Years of protection: 5.4 to 6.6 years

*Stroop Word Test (processing speed):*

Significant improvements in all training groups (p < 0.05)

Effect sizes: d = 0.240 to 0.271

Years of protection: 2.3 to 2.7 years

**Non-significant outcomes:**

Stroop Color Test: No significant effects

Stroop Color-Word Test (interference): No significant effects

Controlled Oral Word Association Test (verbal fluency): No significant effects

Digit Vigilance Test (sustained attention): No significant effects

**Dose-response:**

The 14-hour on-site group (with booster sessions) generally showed the largest effect sizes, suggesting some dose-response relationship

However, the 10-hour at-home group performed similarly to the 10-hour on-site group, indicating that home delivery is feasible and effective

**Age effects:**

Effects were generally similar across the two age bands (50–64 vs. 65+), suggesting the training works for middle-aged as well as older adults

The improvements were small-to-moderate by standard conventions (Cohen's d of 0.2–0.6). To put this in concrete terms:

On the UFOV test, the average improvement in the training groups was equivalent to what would normally be seen in someone 3–4 years younger. If a 70-year-old trained, their processing speed at one year looked like that of a 66–67-year-old who hadn't trained.

On the Symbol Digit Modalities Test, the effect was larger — equivalent to 5.4–6.6 years of protection. This means a 65-year-old who trained performed at the level of a 58–60-year-old on this test.

On Trail Making Test B (a measure of cognitive flexibility), the effect was smaller — equivalent to 1.5–2.0 years of protection.

These effects are roughly comparable to the difference between someone who exercises regularly versus someone who is sedentary, or between someone who gets adequate sleep versus someone who is chronically sleep-deprived. They are meaningful but not transformative — you won't suddenly become a genius, but you might notice that you can follow conversations more easily or react faster while driving.

**Acknowledged by authors:**

The primary outcome (UFOV) is thematically similar to the training task, so "training to the test" cannot be ruled out — the improvements may partially reflect practice on a similar task rather than general cognitive enhancement

The attention control group used crossword puzzles, which may not have matched the training groups for engagement, novelty, or expectations

The study had no no-contact control group, so natural age-related decline over the year could not be measured

The sample was predominantly white and well-educated, limiting generalizability to more diverse populations

The one-year follow-up cannot speak to longer-term durability of effects

**Critical reader observations:**

The lead author (Wolinsky) had a prior consulting arrangement with Posit Science, the company that owns the training software. While the authors state no financial conflict, this relationship could introduce subtle bias in study design or interpretation.

The study was funded by NIH, not the company, which reduces but does not eliminate industry influence concerns.

Only 5 of 9 cognitive tests showed significant effects, and the effects were on tests most similar to the training (processing speed) rather than on tests of higher-level cognition (executive function, verbal fluency).

The "years of protection" metric is derived from cross-sectional age comparisons and assumes linear decline, which may not accurately reflect individual trajectories.

The at-home group had phone support, which may not be available in real-world implementation.

Participants were volunteers from a single medical center, likely more motivated and healthier than the general population.

For someone running their own n=1 experiment:

**What to test:**

The specific intervention: Visual speed of processing training using the "Road Tour" program (or similar commercially available software like BrainHQ's "Double Decision" or "Target Tracker" exercises). The key feature is that it trains divided attention — identifying a central target while simultaneously locating a peripheral target, with progressively shorter display times.

Dose: 10 hours total, delivered in sessions of 30–60 minutes, 2–5 times per week over 2–5 weeks. The study used two-hour sessions, but shorter sessions may be more practical and likely still effective.

**Minimum meaningful duration:**

10 hours of training over 2–5 weeks, with follow-up assessment at 1 month and 3 months post-training. The study found effects at one year, but for a self-experiment, shorter follow-up is reasonable.

If you want to test the booster effect, add 4 additional hours at week 11 and week 35 (2 hours each).

**What to measure (specific metrics):**

**Primary metric:** Processing speed on a divided attention task. You can use the UFOV test if available, or a free alternative like the "Trail Making Test" (available online or through cognitive testing apps). Measure time to completion in seconds.

**Secondary metrics:**

- Trail Making Test B (executive function) — time in seconds

- Symbol Digit Modalities Test (processing speed) — number correct in 90 seconds

- Simple reaction time (available through many apps) — milliseconds

- Self-reported cognitive function: "How often do you lose your train of thought?" or "How quickly can you follow a fast-paced conversation?" on a 1–10 scale

**Real-world transfer:** Driving reaction time (if you have a simulator or can safely test braking response), speed of completing crossword puzzles or Sudoku, time to complete a familiar task like sorting mail

**Key confounds to control for:**

**Practice effects:** Cognitive tests improve with repeated taking. Take baseline measures 2–3 times over a week to establish a stable baseline before starting training.

**Expectation/placebo:** If you know you're doing the "real" training, you may expect to improve. Consider a control period where you do a different cognitively engaging activity (like crossword puzzles or learning a new language) for the same amount of time, then compare.

**Sleep:** Poor sleep impairs cognitive function. Track sleep quality (e.g., using a sleep diary or app) and ensure consistent sleep during both training and control periods.

**Stress and mood:** Stress and depression affect cognitive performance. Track daily stress (1–10 scale) and mood.

**Physical activity:** Exercise improves cognition. Keep exercise consistent throughout the experiment.

**Caffeine and alcohol:** Both affect processing speed. Standardize intake or track and control statistically.

**Time of day:** Cognitive performance varies diurnally. Test at the same time of day for all assessments.

**Other cognitive activities:** If you start doing more puzzles, reading, or learning during the experiment, this could confound results. Keep non-experimental cognitive activities constant.

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

A 10–20% improvement in processing speed on divided attention tasks (e.g., UFOV or Trail Making Test A) from baseline to post-training

A 5–15% improvement on Trail Making Test B (executive function)

A 10–20% improvement on Symbol Digit Modalities Test

These improvements should be larger than any changes seen during a control period (e.g., crossword puzzles)

Real-world improvements might include: feeling quicker in conversations, noticing you react faster while driving, completing routine tasks more efficiently

A negative result (no improvement) would suggest that