Nikolai V. Kharitonov | Moscow, 2026
Abstract
This paper proposes an interdisciplinary model that considers intelligence as a hierarchical dissipative system whose goal is to create entropy gradients — ordering chaos through the formation of stable neural structures. Based on this approach, the phenomenon of addiction is analyzed: it is interpreted not as a chemical breakdown or moral weakness, but as a “bypass neural network” — a stable pathological algorithm formed in childhood as a survival strategy under conditions of an unsolvable problem. The model implies that effective therapy should work not at the lower (chemical) level, but at the higher level — the level of decision-making algorithms. As a promising direction, the method of “virtual good childhood” is proposed: the technological creation of experience that rewrites the traumatic neural network through memory reconsolidation mechanisms.
Keywords: intelligence, dissipative structures, anegentropy, neuroplasticity, addiction, memory reconsolidation, therapy.
Introduction: Intelligence as a Machine of Order
Human intelligence is a phenomenon that can usefully be considered not only in psychological or philosophical categories, but also in the language of exact sciences. This approach allows us to see general patterns where humanitarian knowledge often gets bogged down in describing particular cases.
From the standpoint of thermodynamics and systems theory, intelligence is a hierarchical, open, dissipative structure that converts high-quality energy (chemical) into various types of work: internal (maintaining order) and external (mechanical impact on the environment).
The brain consumes about 20% of the body’s energy while accounting for only 2% of its mass [1]. This energy is spent not on mechanical movement of the skull, but on maintaining a complex structure — neural connections. From an infinite number of possible configurations of matter interaction, intelligence seeks and finds the most effective one, considering the goal and resource expenditure.
The key thesis underlying our further reasoning: the result of intellectual labor is a new neural connection (structure), created through energy dissipation, which stores information and is capable of controlling future actions.
However, one iteration of the result does not guarantee a high level of efficiency. Moreover, under certain conditions, intelligence can form neural connections that lead the system into a dead end. A vivid example is addiction.
Part 1. Physics of Thinking: Basic Concepts
1.1. The Work of Intelligence in Terms of Thermodynamics
In classical physics, mechanical work is defined as A = F · S (force multiplied by displacement). From this point of view, the process of thinking is not work — the head does not move in space under the influence of thoughts.
However, the energy expended during thinking is enormous. It is used for:
- maintaining the electrical potential of neurons;
- synthesis of neurotransmitters;
- operation of ion pumps;
- creation and strengthening of synaptic connections.
All this energy ultimately dissipates — is released as heat. But in the process of dissipation, a structure emerges: a new neural connection, a new pattern, new knowledge.
Dissipation is not a defect, but a condition for the existence of thought. In physics, there is the concept of “order.” The brain expends energy to create ordered structures (thoughts, images, neural ensembles) in conditions of external chaos.
1.2. Hierarchy of Work in an Intellectual System
Considering the system “a person solving a problem,” three types of work can be distinguished:
A. Chemical work (internal). Synthesis of proteins, creation of neurotransmitters, operation of ion pumps. This work maintains the system in working condition (homeostasis).
B. Informational work (computational). Information processing is a process that necessarily requires energy expenditure (according to Landauer’s principle, erasing one bit of information leads to heat dissipation) [5].
C. Mechanical work (external). The ultimate goal of thought: muscles contract, the hand saws a board, a stool is created.
1.3. Efficiency and Coefficient of Performance
The key parameter is the coefficient of performance (COP):
COP = Auseful / Eexpended = 1 – (Qdissipation / Eexpended)
The higher the COP, the less energy is wasted on heating the environment and the more is used to achieve the goal.
Intelligence is unique in that it can increase the overall COP of the system by expending energy on preliminary modeling. It is better to think for an hour and make an ideal stool in 10 minutes than to immediately start sawing and redo it three times. In the first case, intellectual dissipation is higher, but the overall dissipation of the system is lower.
This is anegentropy in action: the conscious, purposeful creation of an opportunity gradient to minimize total energy expenditure [4].
Part 2. How Pathological Neural Networks Are Formed
2.1. The Normal Decision-Making Cycle
In a healthy intellectual system, the decision-making process goes through the following stages:
- Receiving the task — sensory systems register the problem.
- Task processing — primary analysis, classification.
- Search for options — hypothesis generation (the most energy-intensive stage).
- Choice of option — comparison of forecasts, suppression of alternatives.
- Formation of the task for the entire IS — translation of the solution to the executive level.
- Task execution — mechanical work.
- Comparison of result with task — feedback, strengthening of successful neural connections.
2.2. Formation of a “Bypass Neural Network”
Under certain conditions, this cycle fails. A classic case is childhood trauma.
A child receives a task: “I’m scared, I need my mother’s love.” Available solutions are limited: cry, endure, seek approval. If the mother is cold or rejecting, the task is not solved by any effort. The gradient is not relieved, energy is wasted.
In this situation, the brain (possessing high plasticity in childhood) seeks a bypass. And it finds one: temporary relief through primitive chemical self-regulation (thumb sucking, eating, later — alcohol). A bypass neural network for decision-making is formed:
Irritation (unsolved problem) → Chemical stress relief
This path is a short circuit. The signal bypasses the cortex (where analysis should occur) and goes directly to subcortical structures responsible for primitive pleasure/pain relief.
2.3. Why Is This Network Stable?
First, it is formed in childhood, when the brain is most plastic. The connection “approval = life” is deeply embedded — in the brainstem and limbic system, which do not understand words and do not distinguish past from present.
Second, it is energetically cheap. No need to think, analyze, endure — just take a substance, and the tension goes away. The brain is lazy (energy-saving), it chooses the short path.
Third, in adolescence or adulthood, this neural network receives powerful reinforcement: alcohol removes inhibitions, the person finally expresses pent-up feelings to the mother, an illusion of problem-solving arises. The connection is fixed as “working.”
2.4. Hierarchical Breakdown
Normally, the “irritation” signal should trigger a higher process — search for options, analysis, choice. In addiction, a breakdown of the management hierarchy occurs:
| Characteristic | Higher Level (Intellect) | Lower Level (Chemistry) |
|---|---|---|
| Function | Analyzes, chooses, plans | Changes mood, relieves pain |
| Tool | Neural networks | Neurotransmitters |
| Speed | Slow (seconds-days) | Fast (instantaneous) |
| Result | Long-term solution | Short-term symptom relief |
In addiction, the lower level takes control. The higher level is switched off. The process “falls” downward and never reaches a solution to the real problem.
The system’s COP tends to zero. Energy is spent, but no useful work is produced. The task (e.g., “to be loved”) remains unsolved, and the person falls into a trap: each chemical stress relief strengthens the bypass network and weakens the decision-making network.
Part 3. Why Chemical Treatment Doesn’t Work
From the described model follows a simple and harsh conclusion: treating addiction only with chemicals (replacement therapy, receptor blockers) is pointless.
Chemistry works at the lower level. It can temporarily reduce cravings, relieve withdrawal, block the high. But it does not restructure the bypass neural network. It’s like trying to fix a broken algorithm by replacing the processor without reinstalling the program.
Moreover, purely chemical treatment often aggravates the situation. Suppressing symptoms without solving the cause creates an illusion of control, while the bypass network remains and awaits its moment — the first serious stress.
Statistics confirm this: medication-based treatment of addictions has a huge relapse rate (up to 70–90% in the first year). Comprehensive programs that work with thinking and behavior work better because they attempt to restore the higher level.
Part 4. Therapy as Engineering: The Principle of Rewriting
If addiction is a bypass neural network formed in childhood as a way to survive an unsolvable situation, then treatment should be an engineering task of rewriting this neural network.
4.1. Memory Reconsolidation
Modern neuroscience confirms: memory is not a hard drive, but a process. Every time we recall an event, we reassemble the memory anew. At that moment (the reconsolidation window), it is plastic. It can be changed [6].
Traumatic memory is also a memory. The child interpreted the situation as life-threatening (although in reality they would not have died) and encoded it. But this interpretation can be rewritten — if the brain is given a new experience in the same context.
4.2. The “Virtual Good Childhood” Principle
The model implies: for the bypass network to cease being dominant, the brain needs to be provided with an alternative experience where:
- the context is as close as possible to the original (the state of a child seeking love);
- the threat is absent;
- the need is satisfied without additional effort.
This can be achieved through technologically created virtual environments:
- VR therapy immersing a person in childhood scenes with an “ideal mother”;
- generation of photo and video memories using neural networks;
- hypnotic reliving with an altered scenario;
- combination of these methods with a chemical “plasticity window” (psychedelics, MDMA).
The brain, having received such an experience in a state of high suggestibility (close to childhood), can rewrite the old neural network. Not erase it, but render it irrelevant — lay down a new, wider path along which it is more advantageous for the signal to travel.
4.3. Ethical Aspect
Objection: “But these are false memories! This is deception!”
Response within the thermodynamic model: The old neural network itself was false. The child made a mistake, mistaking lack of love for death. They created a map that did not correspond to reality. The new neural network is more adequate: “I am safe, even if not everyone likes me.”
The brain does not need “objective truth.” It needs a working model that allows it to spend less energy on fear and more on life.
What is ethical is not what is “true” in the historical sense. What is ethical is what reduces suffering and increases the system’s COP without destroying it. Replacing a destructive neural network (trauma) with a constructive one (safety) through the creation of targeted experience is not deception, but engineering. It is treating causes, not providing crutches for symptoms.
Part 5. Perspectives and Conclusions
The proposed approach is a special case of a more general principle formulated in the “Manifesto of Anegentropy”: intelligence as a conscious creator of entropy gradients [4].
If addiction is a failure at the highest level of control, then it must be treated at the highest level. Not with pills (chemistry), but by restructuring neural networks through new experience.
This opens the door to fundamentally new therapeutic methods:
- Personalized VR environments simulating a “good childhood” tailored to a specific trauma.
- Neural interfaces monitoring the activation of old neural networks and offering alternative patterns in real time.
- Pharmacological opening of plasticity windows followed by psychotherapeutic or VR rewriting.
- Creation of “healthy memory banks” — libraries of virtual experience available for “implantation.”
Humanity’s main enemy is not other people, but chaos, entropy, the decay of structures. Addiction is just a particular manifestation of this enemy within an individual nervous system. But by understanding the physics of the process, we can begin a war against the enemy, not against its victims.
References
[1] Raichle, M.E., & Gusnard, D.A. (2002). Appraising the brain’s energy budget. Proceedings of the National Academy of Sciences, 99(16), 10237-10239. https://doi.org/10.1073/pnas.172399499
[2] Prigogine, I. (1977). Nobel Lecture: Time, Structure and Fluctuations. https://www.nobelprize.org/prizes/chemistry/1977/prigogine/lecture/
[3] Schrödinger, E. (1944). What is Life? The Physical Aspect of the Living Cell. Cambridge University Press.
[4] Kharitonov, N.V. (2026). Manifesto of Anegentropy: The Basis of a Unified Picture of Reality. https://anegentropy.com
[5] Landauer, R. (1961). Irreversibility and Heat Generation in the Computing Process. IBM Journal of Research and Development, 5(3), 183-191. https://doi.org/10.1147/rd.53.0183
[6] Nader, K., Schafe, G.E., & LeDoux, J.E. (2000). Fear memories require protein synthesis in the amygdala for reconsolidation after retrieval. Nature, 406, 722-726. https://doi.org/10.1038/35021052
The article was prepared based on materials from an extended dialogue with the author of the “Manifesto of Anegentropy,” N.V. Kharitonov.
© Nikolai V. Kharitonov, 2026. This material may be freely used for non-commercial distribution with a mandatory active hyperlink to the original: https://anegentropy.com
Published: March 14, 2026
