Predictive Brain
How Expectations Shape the Perception of reality.
· 15 min read
In the previous text, we explored the question of why the human brain can enter a wide spectrum of altered states of consciousness, from meditation and flow to dissociative states. We demonstrated that from an evolutionary perspective, the key principle might not be a single optimal mode of functioning, but rather the nervous system’s ability to flexibly change its organization according to the current context.
This text builds upon that framework and focuses in more detail on predictive processing—an influential group of theories suggesting that during perception, decision-making, and bodily regulation, the brain does not merely work with signals coming from the environment and the organism, but also with continuously generated expectations.
Predictive processing is not a definitively confirmed, unified theory of the brain. It is a theoretical framework that integrates research findings on perception, attention, learning, motor control, and interoception. Its main contribution lies in its ability to explain how prior experience, current context, and bodily state influence what a person perceives and how they react to a situation.
The Brain as an Active Interpreter, Not a Passive Recorder
In the previous text, we introduced the principles of predictive processing and prediction error on a general level. Here, we explore them in greater detail using perception itself as an example.
The common intuitive view suggests that the senses first record the external world, and the brain subsequently processes the received information. In reality, perception is a much more active process. Sensory signals are often incomplete, variable, and noisy, so the brain must continuously estimate their most likely cause—perception thus arises from the interaction between the information flowing from the senses and prior knowledge, experiences, and contextual expectations.
This process is usually described as hierarchical. Higher levels of processing represent more general relationships and contexts, while lower levels deal with more specific properties of stimuli, such as edges, movement, color, sound intensity, or the position of body parts. The notion that “it is safe in this room” is therefore not a single isolated prediction, but a complex conclusion derived from sensory information, memory, bodily state, emotional appraisal, and knowledge of the social context.
At the same time, the brain has no direct access to reality itself, only to available signals—a prediction error therefore does not mean that it “made a mistake in describing reality,” but rather that a discrepancy arose between the input the system expected and the input it actually received.
The nervous system can resolve such a discrepancy in several ways. It can adjust its expectation, focus attention on further information, change its interpretation of the situation, or act in a way that alters the sensory input itself. For example, if a person mishears a sentence, they might change their estimate of its meaning, ask for it to be repeated, or step closer to the speaker. Prediction and action are therefore not separate processes.
Precision Weighting and the Role of Attention
The brain does not treat all signals as equal. It continuously estimates which pieces of information are reliable in a given situation and which should be given more weight. In predictive processing theories, this process is referred to as precision weighting.
Here, precision does not mean objective truth. Rather, it refers to the estimated reliability or informational value of a specific signal. If the environment is dark and the visual input is unclear, the brain may rely more heavily on prior expectations. Conversely, if the sensory signal is distinct and unambiguous, it can more easily correct the expectation.
Attention is closely linked to precision weighting. In some models, attention can be understood as one of the mechanisms that increases the salience of selected signals. When a person focuses on their breath, pain, another person’s facial expression, or a sound in the room, they change the relative weight of this information in subsequent processing. Attention, however, is not entirely identical to precision weighting. It is one of the processes that can determine which signals will most significantly influence the resulting perception and action.
This principle helps explain why two people can perceive the same thing differently right down to the level of which information their nervous system prioritizes—not just because of differing opinions.
The Automatic Evaluation of Relevance and Safety
The nervous system continuously processes the biological and behavioral relevance of a situation. It evaluates, for example, whether a certain stimulus is related to a threat, a reward, a chance to connect, a need to act, or a need to conserve energy. A large part of these processes occurs without conscious decision-making.
Within Stephen Porges’ Polyvagal Theory, the term neuroception is used for the non-conscious distinction between safety and threat. This term has gained popularity, particularly in psychotherapeutic and popular science circles. However, it is important not to confuse it with a strictly defined and universally confirmed neurobiological mechanism. Some specific anatomical and evolutionary claims of Polyvagal Theory are subject to academic criticism.
A more widely accepted general finding is that the brain and the autonomic nervous system react to biologically significant stimuli without all the intermediate steps having to enter consciousness. Such processing can affect posture, muscle tension, breathing, heart rate, hormonal regulation, attention, and readiness to act.
The resulting meaning of a situation does not stem solely from the current stimulus. It can be schematically expressed as follows:
Meaning of experience = a function of the current stimulus, prior experience, bodily state, and social context.
This is not a mathematical equation intended for precise calculation, but an illustrative expression showing that the same event does not necessarily hold the same meaning for different people.
The physical closeness of another person, for instance, can be associated with trust, support, and calm. For someone else, that exact same closeness might trigger heightened vigilance, especially if it has been linked in the past to unpredictability, rejection, or a loss of control. Such a reaction may not be a conscious choice and cannot simply be considered a character trait. It can reflect learned expectations that automatically influence attention, bodily regulation, and the interpretation of the situation.
However, this does not mean the reaction is immutable or entirely beyond the reach of conscious influence. Automatic expectations can be modulated by current information, regulation of attention, the relationship with the other person, conscious reappraisal, and repeated new experiences.
Predicting One’s Own Body: Interoception and Organism Regulation
Predictive processing does not apply solely to the external world. The nervous system must also continuously estimate and regulate the internal state of the organism. This process is tied to interoception—the processing of signals conveying information about heart rate, breathing, visceral processes, temperature, hunger, thirst, metabolic needs, and other aspects of internal physiology.
At the same time, bodily experience arises through the collaboration of interoception with other systems. Proprioception provides information about the body’s position and movement, nociception about potentially damaging stimuli, and exteroception about the surrounding environment. The brain integrates this information with memory, context, and expectations.
In predictive models of interoception, the brain does not merely receive bodily signals passively. It continuously estimates what state of the organism can be expected and attempts to keep physiological variables within limits compatible with current needs. This principle is related to homeostasis (the maintenance of internal stability) as well as to allostasis (the anticipatory regulation of future demands).
For instance, the organism doesn’t have to wait until there is a complete lack of oxygen during physical activity. Breathing and heart rate change anticipatorily based on the expected exertion. Similarly, the mere thought of a demanding situation can trigger physical preparation before the situation actually occurs.
Interoception and Emotions
The relationship between bodily signals and emotions is the subject of several different theories. Predictive and constructivist approaches emphasize that emotional experience does not arise merely as a hard-wired automatic reaction to an external stimulus. It is also shaped by how the brain interprets the bodily state in light of context, prior experience, and current expectations.
This assertion, however, does not imply that all emotions are simply conscious interpretations of bodily changes. Emotional processes can occur rapidly and without conscious deliberation. Furthermore, there is no consensus among scientists that all emotions arise through a single mechanism. Some theories emphasize evolutionarily prepared affective systems, while others focus on appraisal processes, learning, or the construction of emotional categories.
It is well documented, however, that bodily states influence emotional experience. Similar or partially overlapping physiological changes—such as an elevated heart rate, quickened breathing, and muscle tension—can be experienced in different contexts as anxiety, anger, arousal, or joyful anticipation.
What matters is not just the physical change itself, but its presumed cause. An accelerated heartbeat during a run usually doesn’t prompt the same interpretation as an accelerated heartbeat during a confrontational argument. The meaning of the bodily signal is co-created by the situation, attention, memory, and expectations.
This helps explain why the same person might interpret similar bodily sensations differently at various stages of life. If the context, experience, or beliefs about the meaning of the bodily state change, the resulting emotional experience can change as well.
Prediction Error as a Mechanism of Learning
A mismatch between expectation and outcome can be an important trigger for learning. If an event unfolds exactly as expected, the nervous system usually lacks a strong reason to alter its existing model. If, however, something unexpected occurs, information arises indicating that current assumptions may be insufficient.
The term prediction error is used in several fields and may not always denote the exact same mechanism. In perception, it can mean the difference between expected and received sensory input. In reinforcement learning, it can signify the discrepancy between an expected and an actually obtained reward. In memory research, it can refer to the violation of expectations when retrieving a past experience.
It is therefore inaccurate to assume that every surprise automatically leads to a model change. The outcome depends on the magnitude of the discrepancy, its intelligibility, repetition, emotional significance, the ability to influence the situation, and the current bodily state.
A small and understandable discrepancy can lead to the adjustment of an existing expectation. A massive and hard-to-explain discrepancy, on the other hand, might lead to the conclusion that it is an exception or an entirely new situation. The person may then leave the original model unchanged and instead create a separate explanation.
How specifically the context of threat versus the context of safety affects the outcome of such learning was discussed in detail in the first text: threat does not preclude learning, but rather steers it toward reinforcing and generalizing defensive reactions, whereas a safe and manageable context opens up space for a more flexible updating of the model. Here, let’s just add that the update itself is not dependent solely on the level of safety, but also on the properties of the discrepancy described above—its magnitude, comprehensibility, and repetition—as well as the degree of physiological arousal during which the learning takes place.
For exploration, creativity, and corrective learning, a state where arousal is sufficient to maintain attention, but not so intense as to significantly impair the ability to process new information, is generally more advantageous. Too low an arousal can lead to disinterest or insufficient engagement, while too high an arousal narrows attention to the immediate threat and favors fast, automatic reactions. Between these extremes lies a broad spectrum of states in which the nervous system can compare old expectations with new experiences and update them more flexibly. Safety is therefore not a prerequisite for all learning—however, it can be a crucial condition for those forms of learning that require openness, exploration, tolerance of uncertainty, and the willingness to question existing defensive strategies.
An experience that disrupts an old expectation in a safe, understandable, and manageable way can support a more flexible update. If a person, for example, repeatedly expects rejection but encounters predictable and respectful reactions in relevant situations, a new expectation can gradually form—how exactly such an update occurs is explored in more detail in the following chapter.
Memory Updating Is Not a Simple Overwrite
Popular texts sometimes state that a new experience “overwrites” an old memory. While illustrative, such phrasing can be misleading.
In extinction learning—for instance, during the gradual weakening of fear—the original association often doesn’t vanish. Instead, new learning emerges, indicating that a certain stimulus might not signify danger in that specific context. The old and new expectations can subsequently coexist. Which one manifests depends on the situation, the context, the level of stress, and the time elapsed since the learning occurred.
Under certain conditions, an already stored memory can also be updated during a process known as reconsolidation. When a memory is reactivated, it can temporarily become more susceptible to modification. This process is not automatic, however, and cannot be considered a simple, universal rule. It depends on precise timing, the strength of the original memory, the nature of the new information, and other conditions.
Scientifically, it is therefore more accurate to speak of updating expectations, the emergence of new competing learning, or a reduction in the probability of the original response, rather than the complete erasure or overwriting of the old model.
Why Intellectual Insight Alone Is Sometimes Not Enough
A person might intellectually know that a certain situation is not dangerous, yet still experience a strong physical or emotional reaction to it. This does not mean their conscious knowledge is false or meaningless. Rather, it demonstrates that declarative understanding does not automatically alter all components of a learned response.
Memory and learning are not stored in a single “center”. Declarative knowledge, emotional associations, motor habits, autonomic reactions, and situational expectations utilize partially distinct but interconnected neural systems. Therefore, a person might be able to accurately explain the origin of their fear without their heart rate, muscle tension, or avoidance tendencies instantly changing upon encountering the specific stimulus.
To change an automatic reaction, a new experience might be necessary—one that takes place directly in the situation where the old expectation is activated. Behavioral experiments, exposure, behavioral change, attention regulation, somatic work, and a safe relational context can all play crucial roles here.
This does not mean that verbal psychotherapy works solely through intellectual explanation. Many well-established therapeutic approaches combine dialogue with emotional experience, new behavior, exposure, skills training, and changes in how a person interprets their own bodily reactions. By the same token, one cannot generally claim that somatically focused approaches are always more effective than talk therapies. Individual methods differ in their evidence base, modalities of work, and suitability for specific difficulties.
It is more accurate to state that for certain deeply conditioned reactions, verbal insight alone may not suffice, and that effective change often encompasses multiple levels of experience simultaneously.
Why the Predictive Processing Framework Is Useful
Predictive processing helps explain why a person does not perceive the world as a neutral dataset. Perception is always colored by prior experience, expectations, attention, bodily state, and current goals.
This framework also highlights that the outcome of a surprising experience is not a given—it depends on the context and the characteristics of the discrepancy, as described above. It also helps us understand why rational comprehension of a situation alone sometimes fails to immediately change an automatic bodily or behavioral response.
The predictive framework can also be highly useful in studying altered states of consciousness. Meditation, prolonged sensory stimulation, sleep deprivation, psychoactive substances, intense emotional arousal, or dissociation can—in various ways—alter the relationship between expectations, sensory inputs, attention, and the estimated reliability of individual signals.
Predictive processing, however, does not single-handedly explain every altered state of consciousness. Rather, it offers a common language through which we can formulate hypotheses about why—under certain circumstances—the perception of the body, time, space, self-identity, or the meaning of sensory stimuli shifts.
Conclusion
The brain cannot be accurately described as a passive recorder of external reality. Perception arises from the active integration of incoming signals with prior experience, current context, bodily state, and expectations.
Predictive processing represents an influential theoretical framework asserting that the brain continuously generates estimates about the likely causes of sensory and bodily signals. A discrepancy between expectation and input can lead to a model adjustment, a shift in attention, or an action that alters the situation itself.
This framework allows for the integration of research on perception, interoception, emotions, learning, and organism regulation. At the same time, however, it is crucial not to treat it as a definitively proven unified theory of the brain, nor to use it to unilaterally explain all psychological and therapeutic processes.
Its primary value lies in demonstrating perception as a dynamic process. What a person experiences as reality is not a mere imprint of the surrounding world, but the outcome of an ongoing encounter between what the nervous system expects and what information is currently arriving from the environment and their own body.