10 Reasons Why a Pet That Grows With You Beats a Streak, According to Diet Psychology

Almost every food tracking app runs on the same mechanic: a number that counts up while you comply, and drops to zero the moment you do not. It feels motivating for about eleven days.

The odd thing is that the research on how eating habits actually form does not support that design. It supports something closer to the opposite. Below are ten findings from habit psychology and eating research, and what each one implies about the thing you are looking at when you open the app.

Our bias is on the table: Nommie is built around a creature that reacts to what you eat and grows as you keep going. We think the evidence points that way. Where it does not, or where it is thin, we say so at the end.

1. The logging is not admin. It is the active ingredient.

The most reliable finding in behavioural weight research is also the least glamorous: people who record what they eat lose more weight than people who do not, and the relationship holds across a lot of studies.

A systematic review by Burke and colleagues examined self-monitoring across behavioural weight loss interventions and found a consistent association between adherence to self-monitoring and weight loss outcomes. Later work using mobile tools reached the same place from a different direction.

This matters for a boring reason. If tracking is the treatment, then anything that makes you stop tracking is not a missed data point. It is a discontinued treatment.

2. It takes about 66 days, not 21.

The "21 days to build a habit" line has no serious evidence behind it. It comes from a plastic surgeon's observation about patients adjusting to their new appearance, not from a study of habit formation.

When Lally and colleagues at University College London actually measured it, following 96 people forming new daily habits over 12 weeks, the median time to reach near-automatic behaviour was 66 days. The range was 18 to 254 days.

So the question is not whether you can white knuckle three weeks. It is whether anything in your app is still interesting on day 60.

3. Missing one day does not break the habit.

This is the single most useful finding in the Lally study and almost nobody designs around it. Missing one opportunity to perform the behaviour did not materially affect the habit formation process. The curve absorbed it.

Now consider what a streak counter tells you when you miss a day. It tells you that 43 days of work is gone. That message is not a neutral simplification of the research. It is the reverse of what the research found.

4. Feeling like you blew it makes you eat more.

Here is why that false message is expensive rather than merely inaccurate.

Polivy and Herman described what they called the "what the hell" effect: once a dieter believes they have broken their own rule, they do not return to baseline. They escalate. The cookie becomes the packet. Their boundary model describes how self imposed restraint, once crossed, stops governing behaviour at all for the rest of that day.

The lapse is small. The story you tell yourself about the lapse is what does the damage. An app that dramatises the lapse is supplying the story.

5. Which makes reset-to-zero the wrong mechanic, specifically.

Put points 3 and 4 together and you get a fairly precise indictment.

The habit research says one missed day is close to irrelevant. The eating research says believing you have failed triggers the exact overeating you were trying to avoid. A streak that resets to zero takes an irrelevant event and converts it into a failure signal, at the moment the person is most vulnerable to that signal.

It is not that streaks are unmotivating. It is that they are motivating in the wrong direction at the worst possible time.

6. A creature whose condition depends on yours makes the consequence visible.

Calories are abstract. A number being slightly higher than another number does not feel like anything.

Sun Joo Ahn's group at the University of Georgia tested this directly with children aged 7 to 13. A virtual dog's health visibly improved or deteriorated based on the child's real world fruit and vegetable intake. Children in the virtual dog condition chose to be served significantly more fruit and vegetables than children in the other conditions.

The mechanism is not magic. It converts an abstract quantity into a visible state change in something you care about.

7. Virtual companions have moved real health behaviour, not just engagement.

Engagement metrics are easy to move and easy to fake. Actual behaviour is harder.

A University of Georgia team including Kyle Johnsen and Sun Joo Ahn ran a study with 61 children aged 9 to 12, published in IEEE Transactions on Visualization and Computer Graphics in 2014. Children who interacted with a virtual pet averaged 1.09 hours more physical activity per day than the control group.

That is a real world behavioural outcome from a virtual relationship, which is the claim that actually needs evidence.

8. Looking after something else is easier to sustain than discipline toward yourself.

Reviews of health games repeatedly land on nurturance as a mechanic that works: bringing a character back to health, or keeping it well, gives people a reason to act that self directed willpower does not.

One framing from this literature is caring for the companion as a form of self care. You are not negotiating with yourself about whether you deserve a snack. You are answering a simpler question about someone else, and the answer happens to be the same one.

Anyone who has taken a dog out in the rain on a day they would not have walked themselves already understands this.

9. Autonomy sustains behaviour. Pressure does not.

Self determination theory, developed by Ryan and Deci, holds that motivation lasts when three needs are met: autonomy, competence and relatedness.

The health evidence is reasonably good. Autonomous motivation, meaning acting from personal choice rather than obligation, has predicted exercise adherence and retention in weight loss programmes, with effects persisting past a year. A meta analysis of self determination theory informed interventions in the health domain found small to medium effects on autonomous motivation that held at follow up.

A streak is a controlled motivator. It works by obligation and by the threat of loss. That is the category the research says decays.

10. Relatedness is a psychological need, and a spreadsheet does not supply it.

The third pillar of self determination theory is the one food trackers ignore completely.

Competence, most apps attempt. Autonomy, some manage. Relatedness, the sense of being connected to something, is treated as irrelevant to nutrition software, on the apparent theory that eating is an accounting problem.

It is not. It is one of the most social things humans do. A companion that reacts to your choices is not decoration on top of the tracking. It is the part addressing the need the spreadsheet leaves open.

What this research does not say

A post claiming to be grounded in research owes you the limits, so here they are.

Most virtual pet research is in children. The Ahn and Johnsen studies are children and adolescents. Whether a 34 year old responds to a virtual creature the way an 11 year old does is genuinely not established, and we should not pretend otherwise.

The samples are small. 61 children. 68 children. 96 adults in the Lally study. These are real experiments, not anecdotes, but they are not large trials.

One of the results is a partial null. In the fruit and vegetable study, children with the virtual dog chose significantly more produce, but did not actually consume significantly more than the computer only group. Choosing and eating came apart. That is a real limitation and we would rather you heard it from us.

Self monitoring evidence is correlational in places. People who log more lose more weight. Some of that is the logging working, and some is that people who are doing well are more likely to keep logging.

What the evidence does support, fairly clearly, is narrower than a slogan: tracking works, it needs to survive about two months, one missed day does not matter, believing you failed makes things worse, and a companion whose state reflects your behaviour changes what people do in controlled experiments.

That is the design brief we built Nommie against. Your Nommie reacts to what you actually eat and grows as you keep going. It does not reset to zero because you had a bad Tuesday, because a bad Tuesday is not what the research says undoes you.

The story you tell yourself about that Tuesday is.