Fifteen years ago, Erik and Jim met. Erik had recently completed graduate school at the University of Iowa. Jim was teaching software development. They became friends at once, and soon realized they shared a question that no one else seemed to be asking. How was it possible for writers to convey emotional experiences through words alone?

Asked another way, how were people able to project feelings with words, and how could others detect those emotions and have a genuine emotional reaction simply by reading or hearing them. Simply put, it was a “project and detect” problem.

Of course, there was nothing simple about it, and they soon began researching everything from Channing’s Information Theory to modern neuroscience literature to find out what was known. The search soon involved evolutonary biology, anthropology, psychology, sociology, and linguistics. From there, they began to piece together ideas on how emotions are made, experienced, communicated, received, and reexperienced. Between themselves, they called this part of neuro-aestethics "language dependent neuro-affective simulation," not exactly a memorable phrase, but then, they were a pair of professors.

Over the course of time, they began to consider how best to make use of their work. Should they publish a book? Send articles to journals? Submit their ideas to conferences? In the end, it was Erik who suggested they start a business. Businesses moved quickly and actually made immediate use of research and new ideas. Cutting edge businesses often changed the world in the process. Given that the pair did not want to simply talk about their work or worse, see it filed away in publications almost no one read, they decided to incorporate a business.

At Erik's suggestion, they entered and won a business plan competiton, initially involving educational technology, but the idea did not seem quite right. Jim was interested in creating emotional coaching tools for people with special needs and businesses, but even there, the tools were still missing something. Their neuroscience-based approach to detecting emotions and interacting with users was theoretically sound, being based on the modular architecture of biological brains, but creating outputs algorithmically was too cumbersome. They continued research, with Jim focused on the science and technology and Erik focused on the startup and venture world.

Two things changed everything. COVID19 pushed Erik and Jim into temporarily teaching online, giving them a great deal of live experience with online communication, especially with younger people. This made many people aware of the difficulties many people were having with text-only communication.

The second major change was the public release of large language models. Suddenly, people were no longer sceptical of AI solutions, and the llms themselves offered a fundamentally better way of generating output for users. It also meant that people in the industry began to be open to the possibility of emotional AI.

The llms were not, in themselves, a magic bullet for all use cases. Most llms were large, expensive, and required massive amounts of energy and data to train. That is because they are essentially using an approach that makes them try to stochastically discover and generate their responses all at once.

Natural intelligence does not do this. Instead, it uses a set of interconnected modular brain regions that both cooperate and compete to build up answers and behaviors in a bottom-up way. In fact, emotion often provides the impulse to speak at all, meaning that words do not simply convey feelings, they probably evolved to add information and meaning to them.

Other useful technical advances soon followed. Text-to-speech and speech recognition swiftly improved and soon, it was possible to build a working model using cloud resources of the approach that Jim and Erik had envisioned. Today, their focus is on migrating to edge devices.

In August of 2023, they were able to demonstrate their approach at an event hosted at the SRI Research Center in Menlo Park. This past April, they were selected for Batch 18 of Berkeley's SkyDeck Pad13 Incubator to help launch their startup, Unseen Technologies. They are now taking business and technical steps to move their vision forward.