Владислав Педдер – The Experience of the Tragic (страница 5)
However, theories closely aligned with predictive coding began to emerge more actively only in the late 20th and early 21st centuries. A key role in this development was played by studies on neuroplasticity and the adaptive mechanisms of the brain. Neurobiological research, including investigations into neurotransmitters such as dopamine and the functioning of neural networks, led to significant insights into how the brain uses prediction and internal models to perceive the external world. Foundational figures in predictive coding theory, such as Carl Friedrich von Weizsäcker and Gregory Hopper, proposed that the brain constantly generates hypotheses about the future based on past experience and correlates them with incoming sensory information.
Bayes’ theorem, proposed by the English mathematician Thomas Bayes in the 18th century, became a crucial mathematical tool for analyzing and updating probabilistic hypotheses in light of new data. The core of the theorem lies in its ability to recalculate the probability of a hypothesis based on the arrival of new information. Bayes’ theorem describes how belief (or the probability of a hypothesis) is updated in response to new evidence. In the context of brain function, the theorem can be used to explain how neural networks revise their predictions about the future by integrating both prior and newly acquired experience.
Within the framework of predictive coding theory, this theorem and its formula illustrate how the brain updates its hypotheses (or predictions) about the world based on new sensory data. When the brain encounters novel events (data), it revises its prior probability (prediction) to incorporate this information, thereby enhancing the accuracy of future predictions.
Thus, this process reflects the key feature of predictive coding: the brain does not simply react to data, but actively revises its expectations based on new inputs, always striving to minimize prediction error.
The application of Bayes’ theorem to neurobiology and cognitive science became feasible in the 1980s, when scientists began to understand how the brain might employ probabilistic methods to address problems of uncertainty. In this paradigm, the brain is conceived as a “Bayesian inferencer,” one that generates hypotheses about the world and updates them in response to sensory input using probabilistic principles. The Bayesian model assumes that the brain maintains probabilistic models of future events and adjusts them based on prediction errors – an idea that is directly linked to predictive coding theory.
This updating of probabilistic hypotheses is of fundamental importance, as it enables the brain not only to adapt to changes in the external environment but also to take into account uncertainty in the world, even when information is incomplete. In this sense, Bayes’ theorem and its applications have become fundamental for understanding how the brain, when confronted with uncertainty, can improve its predictions and forecast future events based on prior knowledge.
In summary, the connection between predictive coding theory and Bayes’ theorem has become a cornerstone in the development of neurobiological models that explain how the brain information processes and employs probabilistic computations to anticipate future states. Bayesian theory, as a foundation for managing uncertainty and adaptation, has provided a critical mathematical and cognitive instrument for understanding how the brain functions in a world of constant variability and unpredictability.
Predictive Coding as an Adaptive Mechanism
At the core of predictive coding theory lies the principle that the brain not only reacts to external stimuli but actively predicts them using existing models of the world. The brain formulates hypotheses about what will happen in the future and compares these predictions with current sensory information. When predictions align with reality, prediction error is minimized, allowing the brain to efficiently allocate its resources. However, if an error arises – a discrepancy between prediction and reality – the brain updates its models of the world, which facilitates improved perception and adaptation.
This approach enables the brain to conserve energy and effort by minimizing the need to process all incoming information exhaustively. Instead of interpreting sensory data anew each time, the brain operates with simplified models that are continuously updated based on new sensory inputs. This significantly accelerates information processing and reduces energetic costs. For example, when a person walks down the street, the brain does not analyze every individual step but rather relies on its predictions about what should occur in the next moment.
Predictive coding operates across multiple hierarchical levels, ranging from simple sensory signals (such as sounds or colors) to complex social and interactions abstract ideas. At lower levels, the brain predicts basic sensory features such as shapes and movements; at higher levels, it anticipates more complex phenomena, for instance, people’s intentions or social interaction scenarios.
The Role of Hormones, Neurotransmitters, and the Microbiota in Prediction
The efficacy of predictive coding mechanisms also depends on a multitude of external and internal factors. Hormones, neurotransmitters, gut microbiota, and trauma can significantly influence the brain’s capacity for prediction and adaptation.
Cortisol, the stress hormone, can impair the brain’s ability to adjust its predictions. For example, elevated cortisol levels may disrupt the process of updating the world model, leading to persistent perceptual errors and heightened anxiety. Neurotransmitters such as dopamine play a key role in reward and motivation processes, as well as in the amplification or attenuation of specific brain predictions. Recent research has also demonstrated that gut microbiota can affect cognitive functions and even the brain’s predictive capacities, as microbes interact with the central nervous system, influencing mood and perception.
Trauma – particularly brain injury – can disrupt the neurobiological processes underlying prediction, resulting in cognitive and emotional disorders. For instance, depression and anxiety disorders may be associated with impairments in predictive coding mechanisms, wherein the brain fails to effectively update its world models.
Contemporary brain research indicates that the mind actively constructs and updates models of the world by employing predictive coding and Bayesian approaches.
Predictive coding is a process in which the brain generates hypotheses about what it expects to perceive and compares these hypotheses with actual sensory information. When predictive coding produces a mismatch between the brain’s expectation and sensory input (prediction error), the brain can either update its world model or attempt to interpret the data within existing hypotheses. If the prediction error is excessively large, the brain may sometimes perceive the error itself as reality, which can lead to hallucinations. For example, under conditions of sensory deprivation, when sensory information is insufficient, the brain’s predictions may dominate, giving rise to visual or auditory images that compensate for the lack of real stimuli. In cases of excessive prediction activation, such as during stress or neurochemical imbalance (for example, dopamine excess), the brain may ignore real sensory data and impose its own interpretation. This partially explains hallucinations observed in schizophrenia.
Levels of Predictive Coding:
Low Level (Sensory): The brain predicts simple sensory signals (for example, lines, colors, or sounds). For instance, if you hear the noise of footsteps, your brain predicts that you will see a person.
Intermediate Level (Perceptual): Predictions at this level include more complex structures – images, spoken words, or objects. For example, upon seeing quick movement in the bushes, you infer that it is an animal.
High Level (Cognitive): At this level, the brain forms complex hypotheses, including social interactions and abstract ideas. For example, based on a person’s behavior, you might predict their intentions.
The hierarchy of information processing is based on two types of signals:
Top-Down Predictions: At each level of the brain, predictions are generated about sensory data expected at lower levels. For example, if a higher level predicts that a person is seeing a face, then lower levels will anticipate facial features (eyes, nose, mouth).
Bottom-Up Prediction Errors: When actual sensory input does not match the prediction, a prediction error signal arises. This signal is transmitted to higher levels for model correction and refinement of predictions..
How Does the Brain Correct Errors?
This process occurs through cyclical feedback:
Prediction: A higher level generates a prediction and sends it down the hierarchy.
Comparison: At a lower level, this prediction is compared with the actual sensory input.