Optical Illusion, Self-Supervised Learning, Plato, and Dopamine
(Translated from Chinese by ChatGPT)
Optical illusions are an issue that only a few scholars in specific fields care about and usually do not cause any deviations in daily life observations. However, an intriguing hypothesis emerges when people attempt to explain optical illusions: the human brain does not merely accept external information (such as vision) as is, but rather it is generative. It starts from its own existing cognition of the world to construct an internal representation to simulate external entities. Optical illusions occur because the probability of encountering the true form of these entities in human cognition is very low, while similar illusions frequently occur in daily life. Thus, the brain informs people that what they see is that illusion. Therefore, what people see might not necessarily be real and also reflects their own inner ideas and worldview. In other words, people tend to see what they wish to see or what they are accustomed to.
This conjecture is being confirmed by neuroscience research and is also inspiring algorithms in machine learning. Traditional machine learning includes supervised learning and unsupervised learning. In supervised learning, all data is clearly labeled, akin to giving students both a picture and its features; students then strive to match their guesses with the actual features. Unsupervised learning, however, involves unlabeled data, meaning only the pictures are given to students. Although there are no answers, students can compare a large number of pictures to find underlying patterns. Once the patterns are identified, pictures can be classified based on their features.
These two learning methods are at opposite ends of the spectrum: one relies on data (and labels), and the other relies on exploring the geometric structure of the data space. Research on optical illusions suggests another type of learning method that lies between the two, similar to self-supervised learning. In this approach, students first find patterns in some data, then use these patterns to make educated guesses about features. They compare these guessed features with the actual features to make them align as closely as possible, repeating this process continuously to gradually update their understanding of the dataset.
Based on observations, Young Li believes that all human learners are positioned somewhere along this spectrum. Some people need a supervisor or teacher, or external feedback when they learn. Others prefer to figure things out on their own, intuiting hidden patterns in books or establishing connections between new things they encounter and their existing knowledge systems.
Many suggest that the human brain's processing of information is innate. That is, the way a newborn processes and represents external information is already utilizing neural circuits formed during fetal development. These circuits evolve as they grow. This inevitably reminds one of Plato’s theory of reincarnation. In his "Phaedrus," Plato posits that people arrive in this world carrying memories from past lives, harboring perceptions of beauty and goodness in their souls, though these memories lie dormant until something in this world awakens them, similar to the initial meeting of Baoyu and Daiyu. Aside from evolutionary instincts, what else might newborn brains carry?
Plato's view is poetic, yet the existence of optical illusions symbolizes the fog of human sentimentality. Often, as in "The Truman Show," people live within their constructed worlds. Those in pursuit of truth seek to discover reality and see the world as it truly is. Young Li is a highly emotional person, sometimes blending emotion and rationality, and it is often only in heightened emotional states that he can effectively engage in scientific research. This kind of stimulated state is necessary for his mental world to function optimally. If it stagnates, he falls into a low-energy state of memories and emotional fog, where more traumatic memories surface first, leading to restless thoughts. This might also be an illusion, as he subconsciously selects more negative aspects, or interprets events pessimistically. Negative emotions in low-energy states can trigger a chain reaction, making one realize they've delved too deeply into fantasy, far from reality.
Within the illusion, there may be hidden divine beauty and goodness, as well as fogs that need courage to escape. At such times, one yearns for guidance, like Beatrice in Dante's "Divine Comedy" or John Nash's wife in "A Beautiful Mind." Nash's tearful farewell to his hallucinations in the film left a deep impression.
From optical illusions, much can be discussed. But Young Li particularly recalls a recent casual conversation with a friend. He enjoys dialogues with friends in the humanities. During one such conversation, he mentioned feeling his curiosity waning. His imagination is so rich that before even learning something new, he can already imagine the joy of gaining new knowledge and the sensation in his brain, diminishing his motivation to actively learn. His friend responded, "If you're learning for joy, you might as well inject dopamine directly." They shifted topics, but this unexpected reply stuck with him. He realized a clear difference between his perception and his friend's: his life, or more broadly his learning, is largely driven by happiness or some form of satisfaction. For someone driven by sentiment, as imagination and the spiritual world enrich, dopamine secretion becomes more active, making happiness and satisfaction increasingly cheap, leading to a lack of intrinsic motivation. For his friend, pleasure is not everything in life.
He remembers the sparkle in his friend's eyes during a philosophy class, pursuing some form of truth. Perhaps people's pursuit should not be happiness or satisfaction itself (in ancient Greek terms, not honor). These feelings are important, but they are not the end goal. Truth is a long journey with no end, so there's never a day when satisfaction stops one from advancing. Such rational drive prevents weariness, even when stepping into the same river repeatedly. The world of ideas is vast; even Plato's chariot of love likely needs the reins of cool rationality, a sense of mission, and responsibility.
Truth isn't confined to papers, studies, or abstract theories; it also exists in everyday life, in interactions with others, and in emotions and love. For the emotionally driven, not only can one seek the truth of the physical world and the world of ideas, thinking about what is good and just, but also the truth in life, emotions, and love. This reminds him of "Ordinary World," a book he loved as a child, which discussed the significance of labor not as a means to an end but for some form of spiritual "sublimation." If rationality and emotion coexist, maintaining the stimulated state necessary for exploring truth consumes considerable energy, like a rollercoaster. Even with great effort, there will be times of rest and overwhelming inner emotions. Don't avoid or embellish; with the passion for life shown in "Ordinary World," step into your emotions and gracefully step back out. Without a guide like Virgil or Nash's wife, let rationality lead. Open-minded and adaptable, responding freely to circumstances.
May the reader navigate through the fog of emotions, letting their inner spirit lead, freely exploring the truth.