Training with Other People
Training alone is fine until the week you do not feel like it. Somebody tested what happens when the only thing you add to a programme is other people.
The thing worth saying
Training with other people gets filed under nice-to-have. Good for morale, irrelevant to results.
That framing survives because the alternative is hard to measure. You cannot easily separate what the company does from what the training does.
So look at studies that isolated exactly that variable, held everything else constant, and watched what happened to behaviour over weeks.
The two curves
This is the cleanest demonstration in the article.
A mixed-methods pilot feasibility study compared two interventions in older adult care facilities, each over a 4-week baseline period followed by a 4-week intervention period. The first was reminder-only, with ten participants. The second combined the same reminders with avatar-based virtual social interaction, also ten participants. Primary outcomes were daily step count and medication adherence.
The reminder-only group showed a transient increase in step count in week 1 of 16 percent. By week 4 it had declined to 10.3 percent below their own baseline.
Below baseline. The reminders did not just stop working, the group ended up moving less than before the intervention started.
The reminder plus social support group maintained substantial improvement throughout, reaching 58.4 percent above baseline at week 4, with statistically significant differences between groups at all time points. Medication adherence showed similar trends.
The framing in the paper is that conventional reminder-based digital interventions often produce only transient effects. This is what transient looks like plotted against the alternative.
Ten participants per group in a pilot feasibility study among older adults in care facilities, using avatars on a tablet rather than a training partner in a gym. It is small, it is a different population, and it is measuring steps rather than lifts. What it isolates cleanly is the difference that social contact makes when everything else is identical.
Social dose is a real design variable
Somebody has now proposed treating this as a dose, the way we treat volume.
A narrative review of group-based physical activity interventions proposes the concept of social dose to complement the conventional exercise dose. Exercise dose refers to the frequency, intensity and duration of activity. Social dose refers to the structured exposure to opportunities for social connection embedded within the group-based activity.
That is the useful idea. Not that groups feel nicer, but that the amount of social contact in a training setup is a design parameter you can increase or decrease, and that changing it changes the outcome.
The review was theory-guided, synthesising literature across social psychology, exercise science and public health, and organises the evidence around three psychosocial pathways: belongingness, social identity and peer support.
Its starting problem is the giveaway. Group-based physical activity has emerged as a potentially scalable approach for reducing loneliness, but intervention effects are heterogeneous and are not adequately explained by exercise dose alone.
Not adequately explained by exercise dose alone. Two groups doing the same amount of the same exercise get different results, and the leftover is social.
This is a narrative review proposing a framework rather than a trial testing one, and it is oriented toward loneliness as an outcome rather than strength or hypertrophy.
Why this is not just about being more motivated
The obvious counter is that people who show up are simply more motivated, and the group is incidental. There is a study that pushes against that.
A 26-week home-based structured exercise programme in 88 older adults, averaging 67 years, took a person-centred approach to whether psychological profiles predict adherence. Latent profile analysis built the profiles from autonomous motivation, controlled motivation and exercise self-efficacy, and two distinct profiles emerged.
Critically, adherence was measured objectively rather than by self-report: compliance with prescribed exercise volume, quantified as cumulative metabolic equivalent minutes derived from heart rate data reflecting the intensity and duration of every session.
The paper's title states the outcome: similar exercise adherence and physical fitness outcomes were observed across distinct motivation profiles.
Similar. Two genuinely different psychological profiles, one autonomous and highly confident, produced comparable objective adherence and comparable fitness outcomes in a home-based programme.
So the internal state people obsess over did not separate the outcomes here, which is an argument for paying attention to the structure around the training instead. Who you train with is structure. Motivation profile is not something you can rearrange on a Tuesday.
Eighty-eight older adults in a home-based programme over 26 weeks. It does not say motivation is irrelevant, and a different population or programme design could show something else.
What to actually do
Get one person involved, even loosely. The step count study's only difference between groups was social interaction layered onto identical reminders, and the reminder-only group finished below where it started.
Treat the social side as a dose you can turn up. More structured contact rather than incidental proximity: same time, same days, an actual expectation that you will both be there.
Do not wait to feel motivated enough to need less support. In a 26-week programme, two different motivation profiles produced similar objective adherence, so the internal state is a weaker lever than it feels.
Proximity is not the same as connection. That review distinguishes structured exposure to opportunities for connection from simply being in the same room, which is the difference between a gym full of strangers and training with someone.
And if you train alone and it works, keep going. None of this says solo training fails. It says the social component is doing measurable work, so removing it costs you something you should account for.
Those practical points are my reading of the findings rather than tested prescriptions.
Company is not decoration. A pilot study comparing identical reminder programmes found the reminder-only group peaked at 16 percent above baseline steps in week 1 then fell to 10.3 percent below baseline by week 4, while the group that also got social interaction held 58.4 percent above baseline, significant at every time point. A review of group activity proposes social dose as a design variable alongside exercise dose, precisely because outcomes are not adequately explained by exercise dose alone. And in a 26-week programme measuring adherence objectively from heart rate data, two distinct motivation profiles produced similar adherence, which makes the structure around your training a stronger lever than your internal state.
Limits of this evidence
- The step count study is a pilot feasibility study with ten participants per group in older adult care facilities, which is very small.
- It used tablet-based avatars for social interaction and measured steps and medication adherence, not resistance training with a partner, so the mechanism transfers better than the setting does.
- The social dose review is a narrative synthesis proposing a framework rather than a trial testing it, and targets loneliness as its primary outcome rather than training results.
- Its conclusions rest on the literature it selected, with the emphasis and organisation chosen by its authors.
- The motivation profile study included 88 older adults in a home-based programme, so its finding of similar adherence across profiles may not hold in gym-based training or in younger populations.
- That study reports adherence and fitness outcomes across profiles; it did not test adding social support as an intervention.
- No source here directly compared training with a partner against training alone in a resistance training population, so the central argument is assembled from adjacent evidence rather than tested head to head.
- Each claim rests on a single source with no independent replication checked. Treat the piece as provisional.
Sources (3)
- A pilot feasibility study of a tablet-based virtual community application with shared avatars for promoting health behavior change in older adult care facilities. Frontiers in Digital Health, 2026
- From exercise dose to social dose: a narrative review of the components, mechanisms, and design principles of group-based physical activity interventions to reduce loneliness. Frontiers in Psychology, 2026
- Similar exercise adherence and physical fitness outcomes are observed across distinct motivation profiles in older adults participating in a home-based structured exercise programme. European Review of Aging and Physical Activity, 2026