Effects of Physical-Cognitive Dual Task Training on Executive Function and Gait Performance in Older Adults: A Randomized Controlled Trial.

The researchers compared two types of group exercise classes for older adults over 12 weeks:

**Physical-cognitive dual task (DT) training:** Participants performed physical exercises (walking, stepping, balancing) while simultaneously doing cognitive tasks that challenged executive function — such as counting backwards, reciting alternating sequences, or responding to changing rules. The cognitive demands were designed to specifically engage inhibition and working memory.

**Physical single task (ST) training:** Participants performed the same physical exercises but without concurrent cognitive demands. The focus was purely on motor skill execution.

Both groups exercised twice per week for 12 weeks (24 sessions total), with sessions lasting approximately 60 minutes. The key question was whether adding cognitive demands to physical training would produce greater benefits for executive function (specifically inhibition and working memory) and gait performance compared to physical training alone.

The study measured:

1. **Executive function** — using the Random Number Generation (RNG) task, which yields separate indices for inhibition (ability to suppress habitual responses) and working memory updating (ability to track and update information)

2. **Gait performance** — walking speed, stride length, and stride time under three conditions:

- Walking on flat ground (single task)

- Walking while performing a cognitive task (dual task)

- Walking while negotiating hurdles of 6 cm and 30 cm height (with and without a concurrent cognitive task)

**36 healthy, active older adults** (mean age 72.3 ± 5.8 years, range 65–80)

4 men, 32 women

All were physically active, participating in structured exercise classes at a senior leisure centre in Rome no more than twice per week for at least 4 years prior to the study

Exclusion criteria: uncontrolled cardiac illness, metabolic disease, history of cerebrovascular disease, or other pathological conditions that could affect study outcomes or make exercise unsafe

Originally 50 participants were enrolled; 14 dropped out (28% dropout rate) due to illness, non-exercise-related pain, scheduled surgery, or partner sickness

**Executive function — Random Number Generation (RNG) task:**

Participants were asked to say numbers from 1 to 9 in a random order, paced at 40 beats per minute by a metronome, for a sequence of 100 numbers. The randomness of the generated sequence was analysed using six indices:

**Inhibition indices:** Turning Point Index (TPI — higher = better), Adjacency (ADJ — lower = better), Runs (lower = better). These measure the ability to suppress habitual counting patterns (e.g., not saying "1,2,3" or "9,8,7").

**Working memory indices:** Redundancy (lower = better), Coupon (lower = better), Mean Repetition Gap (higher = better). These measure the ability to track which numbers have been used and avoid repeating them too frequently.

Scores were z-standardised and averaged to create two composite scores: one for inhibition, one for working memory.

**Gait performance — Optojump Next photocell system:**

Participants walked at their habitual speed on a 10-metre path between optical bars that measured:

Gait speed (m/s)

Stride length (m)

Stride time (s)

Walking was tested in four conditions:

1. Flat walking (no obstacles, no cognitive task)

2. Flat walking while performing a cognitive task (dual task)

3. Walking while stepping over two hurdles (6 cm and 30 cm)

4. Walking while stepping over hurdles AND performing a cognitive task (dual task + obstacles)

The first and last metres of the walkway were excluded to remove acceleration and deceleration phases.

**Study design:** Randomised controlled trial (RCT) with two parallel groups.

**Randomisation:** Participants were assigned to groups through stratified random sampling, with stratification based on age and general functional ability (as judged by their instructor of the previous 4 years). This is a weaker form of randomisation than simple random allocation because it relies on subjective judgement for one stratification variable.

**Blinding:** The paper does not explicitly state whether assessors were blinded to group allocation. This is a significant methodological limitation — if the testers knew which group participants were in, it could bias the administration or scoring of the RNG task and gait measurements. Participants obviously knew which type of training they received (no sham control).

**Duration:** 12 weeks of training, 2 sessions per week (24 sessions total). Testing occurred before and after the intervention. No follow-up period was included, so we don't know if effects persisted after training stopped.

**Statistical approach:** The researchers used mixed-design ANOVAs (group × time) to compare changes between groups, plus correlation analyses to examine relationships between changes in executive function and changes in gait performance. Effect sizes (partial eta-squared, η²p) were reported.

**What this design can and cannot prove:**

**Can prove:** That the two training protocols produced different changes in executive function and gait over 12 weeks, assuming the randomisation successfully balanced groups at baseline. The RCT design controls for many confounds (e.g., natural ageing, practice effects, placebo effects) if properly implemented.

**Cannot prove:** That the cognitive demands specifically caused the differences — because the DT group received more total cognitive stimulation, but also potentially more cognitive challenge during testing (they were more familiar with dual-tasking). The design cannot separate the specific effect of dual-task training from general increased cognitive engagement or novelty. It also cannot prove long-term benefits or real-world transfer beyond the laboratory tasks.

**Major weaknesses:**

- Small sample size (n=36 total, only 4 men)

- High dropout rate (28%) — and dropouts were not evenly distributed (8 from ST group vs 4 from DT group, plus 1 discontinuation in each)

- No blinding of assessors (likely)

- No active control for attention/time — both groups got exercise, but the ST group may have felt less engaged

- The stratification method is questionable (subjective instructor rating)

- Only one measure of executive function (RNG task), which is unusual and may not capture all aspects of executive function

- The groups were not perfectly balanced at baseline (ST group was slightly older: 73.7 vs 71.5 years)

**Primary outcome — Executive function (inhibition):**

The ST training group showed a **decrease** in inhibition performance from pre to post (mean composite inhibition score declined)

The DT training group showed a **slight increase** in inhibition performance

The group × time interaction was statistically significant (p < 0.05), meaning the two groups changed differently

Effect size: partial η² = 0.12 (medium effect)

No significant changes were found for working memory updating in either group

**Secondary outcome — Gait performance:**

Both groups improved gait speed, stride length, and stride time under all walking conditions (flat, hurdles, with/without cognitive task) — all p < 0.05

No significant group × time interactions for gait parameters — meaning both types of training improved walking equally

**Correlation between executive function and gait:**

Changes in inhibition performance were significantly correlated with changes in dual-task walking performance, but **only in the DT training group**

This correlation was stronger for the more complex walking condition (with hurdles) compared to flat walking

Specifically: in the DT group, participants who improved more on inhibition also showed greater improvements in gait speed during dual-task walking with hurdles (r = 0.48, p < 0.05)

**Dual-task costs:**

Both groups showed reduced dual-task costs (the performance decrement when adding a cognitive task to walking) after training, but the DT group showed a trend toward greater reduction, particularly for the hurdling condition

**Inhibition decline in ST group:** The ST group's inhibition composite score dropped by approximately 0.3 standard deviations — equivalent to losing about 3–5 years of cognitive function in a healthy older adult (based on typical age-related decline rates)

**Inhibition improvement in DT group:** The DT group's inhibition composite score increased by approximately 0.2 standard deviations — a small but meaningful improvement, roughly equivalent to reversing 2–3 years of age-related decline

**Gait improvements:** Both groups improved walking speed by about 0.05–0.08 m/s on flat ground. For context, a 0.1 m/s improvement in gait speed is considered clinically meaningful in older adults — so these improvements were modest but in the right direction

**Correlation strength:** The r = 0.48 correlation between inhibition change and dual-task gait change in the DT group is moderate-to-strong. For comparison, this is roughly the same strength as the correlation between height and weight in adults

**Acknowledged by authors:**

Small sample size with predominantly female participants (89% women), limiting generalisability to men

High dropout rate (28%) which may have introduced selection bias — those who completed might have been more motivated or healthier

The RNG task, while validated, is an unusual measure of executive function and may not capture all relevant aspects

No follow-up assessment to determine if effects persisted

**Critical reader observations:**

**No blinding of assessors** — this is a major concern. If testers knew group allocation, they could unconsciously influence results, especially on the RNG task which requires subjective judgement during scoring

**Unequal dropout** — 8 participants dropped from the ST group vs 4 from the DT group (plus 1 discontinuation in each). This suggests the ST training may have been less engaging, and the remaining ST participants may have been a biased sample

**No sham or placebo control** — both groups knew which type of training they received. Expectation effects could explain the DT group's better cognitive outcomes

**Baseline differences** — the ST group was slightly older (73.7 vs 71.5 years) and had a different gender ratio (2 men in each group, but different total N). Small baseline differences in a small sample can substantially affect results

**Single cognitive measure** — relying on one test (RNG) for executive function is risky. The RNG task is sensitive to inhibition but may not capture other executive functions like task-switching or planning

**No measure of real-world function** — the study didn't assess whether improvements translated to fewer falls, better daily functioning, or improved quality of life

**Industry funding** — not explicitly stated, but no declaration of conflicts of interest is provided

**Statistical power** — the study was powered for gait speed as the primary outcome, not for executive function. The cognitive findings may be underpowered

For someone running their own n=1 experiment:

**What to test:**

Compare 12 weeks of walking while doing cognitive tasks (dual-tasking) versus walking while listening to music or podcasts (single-task walking)

For the dual-task condition: while walking, perform tasks that require inhibition and working memory — e.g., count backwards by 7 from 100, recite the alphabet backwards, name animals from A to Z, or do simple arithmetic (e.g., "what is 7 × 8?" every 10 steps)

For the single-task condition: walk at the same pace and duration but without any cognitive challenge

**Minimum meaningful duration:**

12 weeks (based on this study), with at least 2 sessions per week

Each session should be 45–60 minutes

You might see trends earlier (4–6 weeks), but the full effect likely requires 8–12 weeks

Test before starting, at 6 weeks, and at 12 weeks

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

**Cognitive:** Use a simple inhibition test — e.g., the Stroop test (free online versions exist) or a "go/no-go" reaction time task. Measure reaction time and accuracy. Alternatively, use the "random number generation" task yourself: generate 100 numbers from 1–9 as randomly as possible, then count how many times you said consecutive numbers (e.g., 3,4 or 8,7) — fewer consecutive pairs = better inhibition

**Gait:** Measure your walking speed over a fixed distance (e.g., 10 metres) under three conditions:

1. Walking normally (single task)

2. Walking while counting backwards by 7 (dual task)

3. Walking while stepping over obstacles (e.g., low boxes or marked lines on the ground) — both with and without a cognitive task

**Dual-task cost:** Calculate the percentage change in walking speed when adding the cognitive task: (single-task speed − dual-task speed) / single-task speed × 100. A reduction in this cost over time = improvement

**Key confounds to control for:**

**Practice effects:** Cognitive tests improve with repetition. Take baseline measurements at least 3 times on separate days to establish a stable baseline

**Fatigue:** Always test at the same time of day, after similar levels of rest and caffeine intake

**Mood and sleep:** Track your sleep quality and mood daily — both affect cognitive performance. Use a simple 1–10 rating for each

**Physical activity outside training:** Keep your other exercise consistent throughout the experiment. Don't start a new exercise program during the test period

**Cognitive activity outside training:** Don't suddenly start doing crossword puzzles or brain training apps during the experiment — this would confound the results

**Expectation effects:** If possible, don't check your results until the end of the 12 weeks. Better yet, have someone else randomise your training schedule so you don't know which condition you're in

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

Your dual-task walking speed improves by at least 0.05–0.08 m/s (about 5–8 cm per second faster) after 12 weeks of dual-task training

Your dual-task cost decreases by at least 10–15% (e.g., from a 20% slowdown when adding a cognitive task to a 5–10% slowdown)

Your inhibition test (Stroop or go/no-go) shows 5–10% faster reaction times with no loss of accuracy

Your random number generation shows fewer consecutive pairs (e.g., from 15 consecutive pairs per 100 numbers down to 8–10)

Critically, these improvements should be larger than any changes seen during the single-task walking condition — if both conditions improve equally, the benefit is from exercise in general, not from dual-tasking specifically