Psychotherapy faces several limitations related to accessibility, adherence, and the social meanings attached to seeking help. First, it can be expensive and is not equally accessible to people from lower socioeconomic backgrounds. Second, psychotherapy—and many of its associated methods—may feel repetitive or boring to some people and may not be engaging enough to support long-term participation. Finally, stigma around psychotherapy and psychological support still persists. For these reasons, there is a critical need to advance therapeutic tools through technological innovation.
There are well-established methods in clinical psychology that offer promising opportunities to make therapeutic tools accessible to a more diverse population. One such method is Cognitive Bias Modification (CBM).
While CBM may help address accessibility challenges, it still struggles with adherence. Many people find the approach boring and repetitive, and they often do not recommend the intervention to others.
To address the adherence and engagement problem, one line of studies turned to a keyword and an associated method called gamification.
Gamification is a term used to refer to a set of methods for making serious contexts like a game. While this direction is important, existing efforts tend to fail at opposite ends of a spectrum. On one end, some “gamified” interventions simply add superficial features (e.g., badges, points, or scores) to traditional methods without changing the core experience.
On the other hand, some of these so-called "therapeutic" games become so entertainment-focused that their connection to a structured psychological intervention becomes unclear.
To address this gap, I designed a highly gamified version of the Cognitive Bias Modification paradigm and called it Help Empath.
PHASE 1
Understanding Intrinsic Motivation
I realized that the key question isn’t what kind of game can be therapeutic for people in need. Instead, I began asking what happens when some people genuinely want to play a game—or immerse themselves in a digital experience—almost obsessively. That led me into a period of literature review and a deep dive into the roots of human motivation.
I can summarize what I learned about motivation in three core principles:
-
When people experience autonomy, competence, and relatedness in a given context, their motivation becomes more intrinsic rather than extrinsic.
-
Experiences such as games are most engaging when the level of challenge matches the player’s skill.
-
Rewards, points, and feedback loops can reinforce engagement—but they work best when they build on the first two principles, rather than replace them.
PHASE 2
Understanding Gamification
While I was still struggling to translate my psychological knowledge of human motivation into a therapeutic game, I came across a gamification model developed by a well-known expert in the field.
The Octalysis model suggests that there are core principles that make games engaging in general. These principles can be borrowed from game design and applied to serious contexts as well.
I learned that the most important part of making a game engaging is making it meaningful to the user. An engaging game should include a mission—something like rescuing a character or fighting against a villain—that gives players a reason to care and stay involved. This goes far beyond simply collecting points for a quick dopamine hit; that approach doesn’t work in isolation, and other factors need to be considered as well. By the end of this phase, I felt confident about the elements that make a game experience truly engaging.
PHASE 3
Back To Psychology - Bias Modification Paradigms
Let us understand first what cognitive biases are.
Cognitive biases are systematic tendencies in how we process information. They often happen because we rely on schemas—mental frameworks that help us interpret the world quickly. Schemas can be useful, but they can also distort judgment by pushing us to interpret ambiguous information in ways that match our expectations.
The concepts of schema and cognitive bias are highly related. A schema acts like a “lens” that influences what we notice and how we interpret what we notice.
Let me give you an example to make it clearer. In Stone, Perry, and Darley (1997), participants listened to the same radio broadcast of a college basketball game, but were led to believe the key player was either Black or White. Even though everyone heard identical information, their evaluations shifted in stereotype-consistent ways: the player was rated as more athletic when believed to be Black, and more intelligent/hard-working when believed to be White. This shows how schemas can produce cognitive bias by shaping interpretation, not the actual facts.
Ok, but what does it mean in psychopathology?
Cognitive models of psychopathology propose that many mental health problems are maintained by systematic biases in how people interpret, attend to, and remember information. For example, someone with health anxiety may interpret normal bodily sensations (like a mild headache or faster heartbeat) as signs of serious illness. They may also selectively focus on these sensations and recall illness-related experiences more strongly than neutral or positive experiences. Over time, these patterns increase distress and reinforce the anxiety.
Now we need to know a little bit about how these biases are assessed.
Cognitive biases in interpretation, attention, and memory (recall) are assessed using different methods. I will explain the methods used to assess interpretation bias in detail, and I will briefly describe how the other two biases are typically assessed.
Cognitive Bias Assessment
Interpretation Bias
Homophones
“Knew” and “new” are homophones: they are pronounced the same way but have different origins and meanings. The same idea applies to certain words that can be interpreted in more than one way—for example, as neutral or positive, or as negative and threat-related. One example is “die” and “dye.”
If a participant consistently chooses the negative meaning of a word when hearing an ambiguous pronunciation, this suggests a biased tendency toward negative interpretation.
Homographs
“Tear” has two different pronunciations—/tɪər/ and /teər/. One has a negative meaning, while the other is neutral. In this case, if a participant consistently pronounces ambiguous words in a way that reflects the negative meaning, it suggests a negative interpretation bias.
Ambiguous Scenarios
This method is often considered one of the most valid approaches in the literature. It’s easier to illustrate it with an example than to describe it abstractly.
Imagine you’re getting ready for a run. As you put on your shoes, you notice a lump that makes it hard to slide your foot in. You might tell yourself it’s just a blister from your new shoes, but you might also think it could be an early sign of cancer.
In this method, participants read short scenarios like this and are asked to complete them in a way that typically requires either a negative or a positive/neutral interpretation.
Scrambled Sentence Task
If I asked you to use five of the following six words to form a complete sentence—
Future / Your / Dismal / Very / Looks / Bright—
You could create either:
-
“Your future looks very bright.”
or -
“Your future looks very dismal.”
If a participant repeatedly forms sentences with a negative meaning across many trials, this is considered evidence of a negative interpretation bias.
Ambiguous Facial Expression
Some facial expressions can be ambiguous and interpreted as either pain or happiness. Explaining the most rigorous ways to measure this type of bias is beyond the scope of this report, but there are several established methods for assessing whether a participant systematically interprets ambiguous facial expressions in a negative way.
Attentional Bias & Memory Bias
The most reliable way to assess attentional bias is through eye-tracking, which measures where a person looks when they are presented with pairs of positive and negative stimuli (e.g., words, facial expressions, or images).
Memory bias is typically assessed by measuring how much positive versus negative material a person recalls after being exposed to a mix of both.
![Happy16 [1024x768].bmp](https://static.wixstatic.com/media/16349e_8c13f1a35e6c405eb2e4536aa911fee0~mv2.png/v1/fill/w_174,h_232,al_c,q_85,usm_0.66_1.00_0.01,enc_avif,quality_auto/Happy16%20%5B1024x768%5D_bmp.png)
![Painful16 [1024x768].bmp](https://static.wixstatic.com/media/16349e_5c8443c1a2544dc5950442d642ac4046~mv2.png/v1/fill/w_174,h_232,al_c,q_85,usm_0.66_1.00_0.01,enc_avif,quality_auto/Painful16%20%5B1024x768%5D_bmp.png)
Happy
Painful
Ambigious
Cognitive Bias Modification
In theory, if we redesign the assessment tools described earlier, we can also use them to train individuals to adopt more positive patterns of thinking. The challenge, however, is that there is not yet enough empirical evidence to conclude with certainty that this training leads to meaningful, real-life changes in people’s attitudes and behavior. One major reason is that these experimental tools often look nothing like what people encounter in everyday life. This gap creates both theoretical and clinical limitations for the field—and it is the core problem this product is designed to solve.
PHASE 4
Integration
At this point, I started thinking about my own experiences with games. I remembered my childhood, when I used to play Super Mario, and I also reflected on my experience with the gamified language-learning approach used by Duolingo.
I designed an avatar named Empath to guide the user as they explore a city. The city and each game level are gamified adaptations of a specific cognitive bias modification paradigm.
.png)
I designed the game with clear progression metrics, feedback loops, failure conditions, and a compelling narrative. Help Empath includes multiple levels, each built around a specific type of cognitive bias and its modification. There is much more to the game than what I can share here, but due to confidentiality concerns, I’m unable to describe the full narrative or level progression. If you’d like to learn more, please don’t hesitate to get in touch.
IMPACT
The project has attracted significant attention from both academic and business sectors. There are multiple venture capital investment opportunities to help grow it into a scalable digital health product.
Researchers in both gamification and cognitive bias modification have spoken highly of the project, noting its clear potential to address one of the most fundamental challenges in psychological interventions.
KEY TAKEAWAYS
The most important lesson I learned was the value of iteration. I realized that the best ideas rarely emerge on the first attempt; it’s the cycle of exploring, researching, prototyping, and critically evaluating that makes a product strong enough to present.
The first version I designed was barely different from what already existed in earlier products and research. However, that prototype taught me why a paradigm shift is needed in how bias modification is embedded in a game.
Working hands-on with something similar to past approaches helped me see, firsthand, what might seem effective on paper but still feels unsatisfying as an experience. It also gave me a metacognitive perspective—an understanding of how individual design elements come together to shape the overall experience. All of this reinforced, once again, the importance of iteration through prototyping.
The first version of my gamified cognitive bias modification design wasn’t very different from what had already been done in past research. However, experiencing it directly—and reflecting on where the experience fell short—helped me understand why it didn’t work well and what needed to change in the next iteration.
Through this experience, I also learned advanced prompt engineering techniques and shifted my perspective on how AI tools can be leveraged to achieve outcomes that might otherwise take months or years. That shift ultimately led to the second project I designed: Empath Convo.
