Know Thyself

How does self-knowledge emerge, and can we improve our introspection?

Photo taken by author, artwork by Tina Hrevušová

In today’s job market, producing a polished CV has become almost routine. Most of us understand that creating one is less an exercise in honest self-examination than in personal branding: selecting and presenting traits that fit a desired image. Yet when we face situations in which a more balanced account of ourselves would be valuable, one that addresses our “real” strengths and weaknesses, many of us experience uncertainty. This is especially true for people inclined toward critical self-reflection.

How, then, does self-knowledge emerge? And should we invest substantial effort in “finding our true selves,” as personal coaches and self-development literature often encourage? When large language models (LLMs) became widely available, part of their appeal came from how human-like they appeared. They could answer questions about their own abilities, limitations, and even aspects of their technical origins, giving the impression that they possessed a form of introspection.

Unfortunately, the similarity seems to end there, as contemporary artificial neural networks are not equipped with dedicated mechanisms for self-investigation. When a large language model talks about its own capabilities, limitations, or origins, it does not use a specific procedure. Instead, these responses are generated in exactly the same way as any other statement: by predicting the most probable sequence of words given the prompt and the patterns learned during training.

In practical terms, this means that an LLM can describe itself only to the extent that relevant information was present in its training data or reinforced during its development. Its apparent self-knowledge is therefore not the result of self-observation but of statistical learning or specific prompting. The model sounds introspective because it has learned how humans discuss introspection, not because it possesses a privileged window into its own inner workings.

Yet for certain kinds of self-knowledge, humans may not be so different. We have to learn and remember basic facts about ourselves: our name, address, occupation, and even aspects of our physical appearance, such as our height or eye colour. Much of what we know about our own preferences is similarly stored in memory. Otherwise, answering a simple question such as “What would you like for lunch?” would require us to reconstruct our tastes from scratch each time.

In fact, a substantial part of what we know about our own personality is learned rather than directly discovered through introspection. Family members, friends, teachers, and society at large provide many of the concepts we later use to describe ourselves. If parents or teachers repeatedly describe a child as lazy, shy, creative, or responsible, the child gradually learns not only the meaning of these traits but also their possible relevance to their own behaviour.

Literature, films, and other cultural narratives play a similar role. By identifying with fictional characters, we acquire additional frameworks for interpreting ourselves. We learn how ambition, jealousy, courage, or empathy look like and begin to compare our own frequent actions with these familiar patterns.

Without such external points of reference, recognising and naming many of our personality traits would be extremely difficult. When we think about ourselves, we rely on a shared conceptual vocabulary to interpret and evaluate our behaviour, emotions, and motivations. This vocabulary allows us to organise a multitude of experiences into coherent descriptions of who we are.

The influence of society is particularly strong at this level. The narratives used to explain patterns of behaviour, as well as the values attached to them, are largely provided by the culture in which we live. These patterns are often linked to broader social categories such as gender, ethnicity, age, or social class. As a result, belonging to a particular group may become associated with certain expected personality traits or behavioural tendencies.

Over time, we may come to see ourselves through these socially constructed categories, incorporating their associated expectations into our self-image. Yet our self-knowledge does not rely solely on social feedback. We also evaluate the consequences of our actions and decisions, often without conscious awareness. By continuously comparing expected and actual outcomes, we build impressions of our own abilities and character.

When the results of our efforts exceed expectations, we may experience a sense of achievement and competence. When they fall short, the sense of disappointment or failure could prevail. Such experiences gradually contribute to the way we see ourselves, providing another source of self-knowledge.

At first glance, this may seem like a well-balanced system. External feedback from society is combined with internal feedback from our own experiences, allowing us to adjust our behaviour and maintain a coherent sense of self. Yet this system has important limitations.

The problem is that neither source of feedback is necessarily objective. We may interpret our achievements or failures in a distorted way, especially when we overlook the broader consequences of our actions or selectively focus on evidence that supports a favourable self-image. External feedback is equally imperfect. Other people often evaluate us through the lens of their own experiences, expectations, and biases, attributing qualities to us that we may not actually possess.

Moreover, when we become part of a group that promotes harmful attitudes or behaviours, the social environment can gradually reshape the standards by which we evaluate ourselves. Actions that might once have triggered guilt or moral discomfort can be reframed as justified, necessary, or even virtuous. As a result, people may continue to see themselves as moral individuals while engaging in behaviour that causes significant harm to others.

History provides many examples of this process. Supporters of totalitarian ideologies, including those who embraced the Nazi regime before and during the Second World War, often perceived themselves as ordinary citizens acting in accordance with values that had become normalised within their social environment.

At this point, you might object that most people do not spend much time consciously analysing themselves. Unless we are particularly narcissistic, deeply self-conscious, or struggling with depression, explicit evaluations of our personality tend to occur only when circumstances demand them. Why, then, should we pay so much attention to the mechanisms of introspection? Excessive self-reflection can even appear counterproductive, distracting us from important goals and limiting our performance.

Yet research suggests that thoughts involving ourselves occupy a surprisingly large portion of our mental life. We frequently revisit past experiences, imagine future possibilities, evaluate our decisions, and reflect on our relationships with others. Much of this activity occurs automatically, without deliberate effort.

Neuroscientists have identified a set of interconnected brain regions known as the Default Mode Network (DMN). This network becomes particularly active when our attention is not focused on an external task, which is why it is referred to as the brain’s “default” mode of operation. Although the full range of its functions remains a topic of ongoing research, the DMN is consistently associated with processes such as autobiographical memory, imagining the future, social cognition, and self-referential thought.

These functions make the network especially relevant to the construction of a coherent sense of self. By integrating memories from different periods of our lives, linking them with emotions, and relating them to future goals and expectations, the DMN helps create the continuous narrative through which we can understand who we are. Introspection, from this perspective, is not an occasional mental exercise but an ongoing process deeply embedded in the brain’s everyday activity.

The self-reflective processes associated with the DMN serve an important practical function. They help us estimate our chances of success in future situations and adjust our behaviour accordingly. The ability to evaluate one’s own capabilities is not uniquely human. Researchers have identified networks resembling the Default Mode Network in several non-human species, suggesting that some form of internal modelling may have deep evolutionary roots.

For animals, quickly assessing whether they can escape a predator or win a confrontation can be crucial for survival. Humans encounter such life-or-death situations less frequently, but we constantly face decisions that require us to judge our abilities, knowledge, and prospects.

The accuracy of these judgments matters because self-perception can directly influence performance. Numerous studies have shown that confidence affects not only how we feel but also how effectively we act. Even subtle reminders of our strengths or weaknesses can alter our results, sometimes regardless of whether those beliefs accurately reflect our abilities.

One well-known series of experiments demonstrated that Asian-American women performed better on mathematics tests when their Asian identity was made salient beforehand, activating a positive cultural stereotype associated with mathematical ability. The same individuals performed worse when other social identities linked to negative stereotypes were emphasised instead.

This does not mean that simply believing we will succeed or fail directly determines the outcome. The underlying mechanism is more subtle and is closely related to how the brain allocates its resources. When we unconsciously evaluate a task as exceeding our abilities, the situation may be perceived as threatening rather than challenging. Depending on its emotional significance, this assessment can trigger a stress response and activate elements of the well-known fight-or-flight mode.

In such circumstances, part of our cognitive resources is diverted toward monitoring potential risks, managing anxiety, or preparing for adverse outcomes. As a result, fewer resources remain available for the task itself. Whether we are taking an exam, giving a public presentation, or competing in a sporting event, our performance may suffer not because our abilities are insufficient, but because we are unable to fully access them under pressure.

These findings suggest that our performance is shaped not only by our actual skills but also by the stories we tell ourselves about those skills. Self-perception is therefore far more than a passive description of who we are; it can actively influence what we are capable of achieving.

Considering the ideas discussed throughout this article, it seems that our identity is not a fixed entity waiting to be discovered through introspection. Rather, it is gradually constructed through a lifetime of experiences, memories, social interactions, and self-reflection. Although external influences can be powerful, we are not merely passive recipients of them. We actively participate in shaping our self-image, for better or worse.

From this perspective, the ancient imperative “Know Thyself” acquires a broader meaning. Self-knowledge is not limited to identifying our strengths and weaknesses. It also involves understanding the psychological and neural processes through which our self-narratives are created and maintained. By becoming aware of these mechanisms, we may gain greater influence over the stories we tell ourselves and, consequently, over the choices we make.

In this respect, there is a parallel with artificial neural networks. Just as learning can modify the behaviour and self-presentation of an AI system, our own patterns of thought can be reshaped through experience, practice, and deliberate reflection. This does not mean that we can simply redesign ourselves at will, but it suggests that self-understanding can be a tool for personal growth rather than merely a search for hidden truths.

While the market is full of self-help books whose claims are not always supported by scientific evidence, some researchers have attempted to translate findings from psychology and neuroscience into practical advice. Examples include Emily Falk’s What We Value and James R. Doty’s Mind Magic, both of which explore how our beliefs, values, and mental habits can influence the way we think and act.