Negativity bias
Baumeister et al. (2001)
Negative events take up more space in memory than positive ones. All else being equal, we remember what hurt.
Science and method
Feellow doesn't reinvent psychology. Every mechanism in the app is backed by peer-reviewed research you can find and critique. Here's what we build on, and what we refuse to do.
In brief
Feellow grounds every feature in scientific psychology, not in wellness promises. Its method rests on ecological momentary assessment (EMA), which measures experience as it happens rather than reconstructing it from memory; on representing affective state along two dimensions, valence and arousal, validated since the 1980s; and on the idiographic method, which looks for what is true for one person rather than comparing them to an average. The generative engine transcribes, rephrases and synthesizes the user's own data without ever interpreting or diagnosing. Feellow is not a medical device and does not replace a health professional.
01 · Methodological framework
Ecological Momentary Assessment (EMA) is a methodological paradigm from health psychology. It consists of measuring a person's experience at the moment they live it, in their natural environment. Not retrospectively, not reconstructed in a session. Shiffman, Stone & Hufford (2008), Annual Review of Clinical Psychology. Perski et al. (2022), Health Psychology Review.
In practice, a Feellow check-in takes less than thirty seconds and can be entered several times a day. It's not a long questionnaire, it's a short, precise trace. It's this temporal density that makes correlations visible afterward.
~600
EMA studies published in health psychology.
80%
median adherence on smartphone, according to recent meta-analyses.
< 30s
duration of a Feellow check-in, in line with validated protocols.
02 · Why real-time
When you describe your week in a session, you're not describing your week. You're describing a reconstruction filtered by your memory, your current mood, and three recall biases well identified in cognitive psychology.
Baumeister et al. (2001)
Negative events take up more space in memory than positive ones. All else being equal, we remember what hurt.
Bower (1981)
Your current mood colors your memory of the entire week. If you're feeling bad today, your memory of yesterday will be darker than it actually was.
Kahneman (2011)
We remember the emotional peak and the last moment, not the average. An entire week can be summed up by a bad Friday.
EMA corrects these biases at the source, by capturing experience at the moment it occurs. That's the methodological reason Feellow asks you to enter short and often, rather than long and rarely.
03 · How we represent your state
Rather than asking you to choose an emotion from a fixed list, Feellow invites you to locate your emotional state on two axes: pleasant or unpleasant, calm or activated. This two-dimensional representation has been validated since the 1980s in emotion psychology. It's more accurate than a fixed list, and more accessible than a long questionnaire. Barrett (2004), Lieberman (2007).
We don't compare your data to an average, a benchmark, or a cohort. We look for what's true for you, week after week. The idiographic method consists of studying the individual rather than situating them within a group. Your patterns belong to you, and they resemble no one else.
04 · The three app mechanics
01
Swann (1983, 2012) · Loewenstein (1994) · Roediger & Karpicke (2006)
The app generates hypotheses about you from your own data. You validate or reject them with a swipe. Self-observation becomes an active cognitive act, not a passive journal. Active recall consolidates memory traces better than reading alone.
02
Kahneman, Knetsch & Thaler (1990)
Each past check-in gains value over time. In month 1, an isolated dot. In month 6, the anchor of a correlation. Perceived value increases because it becomes real and tangible, without needing to guilt you into engagement.
03
McAdams (1985, 2001)
Each month, the app generates a short narrative chapter recounting your emotional month. Not a dashboard, a text. Building the self happens through longitudinal storytelling. You read yourself, and you understand yourself differently.
05 · How your Manual is built
Your monthly Manual is generated from your own data. Feellow's generative engine does three things, mechanically, without interpreting:
It helps you write, not think. It formats, it doesn't heal. You can always see where a generated statement comes from, and request deletion of the data that was used to generate it.
06 · Our red lines
Part of our scientific approach is clearly stating what Feellow doesn't do, and will never do. Here are our red lines, enforceable.
Every feature must be justifiable by evidence-based data from psychology research. If an idea is appealing but doesn't hold up against the literature, we don't ship it.
Never "think positive and everything will be fine". Scientific psychology recognizes that all emotions have a function, including uncomfortable ones. We don't teach you to feel better, we help you know yourself better.
Feellow is not a medical device. The generative engine doesn't diagnose and doesn't advise you. If you need clinical help, talk to a mental health professional.
No guilt-inducing streaks, no user rankings, no false urgency. One notification a day, never a follow-up. Well-being is not an engagement metric, it's the reason the app exists.
Feellow is not a medical device and does not replace a health professional. The references cited are publicly accessible peer-reviewed research.
Last updated, 19/05/2026.