Depression
What Causes Depression? New Research Confirms Ancient Wisdom
Digital CBT proves changing thoughts quickly reduces negative feelings.
Updated April 7, 2025 Reviewed by Abigail Fagan
Key points
- In recent research, reducing belief in a negative thought cut negative emotions by 87 percent.
- Modeling shows negative thoughts cause negative feelings—not the other way around.
- The findings support CBT and hint that digital tools may be able to quickly ease depression.
Jeremy Karmel, co-founder of Feeling Great, also contributed to the research and findings discussed here.
Imagine waking up feeling anxious and hopeless. You think “I’ll never be happy. Nothing I do matters.” The more you dwell on these thoughts, the worse you feel. But what’s really going on? Are your thoughts causing your feelings, or is it the other way around?
For centuries, people have debated the cause of emotional suffering. Is it childhood trauma? A chemical imbalance in the brain? A lack of love and connection? Every school of psychology has its own theory—the list goes on, with hundreds of theories but little conclusive proof.
Over 1,700 years ago, the ancient Greek Stoic philosopher, Epictetus, proposed an entirely different theory. He argued that it’s not life events that upset us, but our thoughts about them. This controversial idea became the foundation of cognitive behavioral therapy (CBT), developed by Drs. Albert Ellis and Aaron Beck. CBT suggests that if you change the way you think, you can change the way you feel.
It’s a bold and exciting claim—but is it true? With data from a new mental health app, we set out to join the debate.
Testing the Connection: Do Thoughts Really Drive Feelings?
In our first study, we analyzed data from 290 beta testers of the Feeling Great app. Unlike traditional studies that rely on broad, retrospective surveys (asking how you’ve felt over the past two weeks), our app measured users’ negative thoughts and feelings in the here and now, using new, extremely sensitive scales.
This distinction is crucial. By tracking moment-to-moment changes, we could test whether beliefs in negative thoughts (e.g., I’m a failure) actually cause negative feelings—or whether it’s the other way around.
Here’s the data on first-time users who successfully reduced their belief in a single distressing negative thought to zero. On average, one-third of the users achieved this goal in just 80 minutes. The impact on negative feelings was remarkable.
As you can see in Figure 1, users who reduced their belief in their negative thought to zero experienced an average 87 percent reduction in seven negative emotions during their first session with the app.
But does this result prove that negative thoughts cause negative feelings? No, because it could just as easily be the other way around. Maybe the app changed negative feelings, which then caused the dramatic reduction in negative thoughts.
Can this riddle be solved? Using advanced statistical modeling, we tested four possibilities:
- Negative thoughts cause negative feelings.
- Negative feelings cause negative thoughts.
- Thoughts and feelings influence each other in a rapid circular feedback loop.
- Negative thoughts and feelings do not affect each other. Instead, an unknown third factor simultaneously affects both.
What We Discovered
- Belief in negative thoughts has robust, direct, and rapid causal effects on negative feelings.
- Negative feelings do not significantly cause negative thoughts—suggesting emotions are largely a consequence, not a cause, of our negative thoughts.
- No evidence of a circular feedback loop or a hidden third factor emerged.
In short, our data strongly supports the central premise of CBT—that thoughts shape emotions far more than emotions shape thoughts. If true, this has profound implications for how we approach mental health treatment.
Replicating the Findings: Can We Trust the Data?
A highly acclaimed 2015 study published in Science found that over 50 percent of behavioral science studies fail to replicate. This means that even the most promising results often collapse under scrutiny.
To ensure that our findings were solid, we conducted a second independent study with 1,393 users of an AI-enhanced version of the app. The results? Practically identical.
Interestingly, with the larger sample size, we did detect a small but statistically significant effect of emotions influencing thoughts—but the effect was barely detectable and far weaker than the impact of thoughts on feelings. This suggests that emotions can sometimes influence thoughts, but the effect is minor in comparison to the power of thoughts to shape emotions.
So, was Epictetus right? It seems so.
Real-World Results
Earlier we showed the dramatic changes in negative feelings in users who crushed their negative thought the first time they used the app. But how about those who didn’t completely reduce their belief in their negative thought to zero? As you can see in Figure 2, the degree of emotional improvement in all users correlated precisely with the amount by which their belief in the thought decreased.
The reduction in users’ beliefs in negative thoughts was strongly correlated with reductions in all seven negative emotions. This strongly suggests that reducing belief in distorted negative thoughts may be one of the fastest and most effective methods for relieving not only depression but a wide variety of negative feelings.
The Future of Mental Health: Human and AI Interventions
These findings open exciting new doors for mental health treatment. The data also suggest that many may be able to modify negative feelings rapidly, affordably, and without any human interventions. The rapid changes we observed suggest that many people can find relief from feelings of hopelessness and despair in hours, or days, without years of therapy.
The data is promising, but we also recognize that human connection and support will always remain essential. For many, face-to-face therapy will continue to be life-changing—and, in some cases, life-saving.
But we have scientific proof that changing how you think really can change how you feel—and that this transformation is possible using an automated digital device.
And that’s a discovery Epictetus would have been proud of.
And how about Freud? He might not be quite as happy.
Facebook image: Nataliya Dmytrenko/Shutterstock
LinkedIn image: DG FotoStock/Shutterstock
References
Open Science Collaboration. (2015). Estimating the reproducibility of psychological science. Science, 349(6251), aac4716. DOI: 10.1126/science.aac4716

