Относится к сборнику статей теори МВАП https://t.me/thinking_cycles
Fundamental structure of response chains
All reaction chains are organized as a sequence of links, each triggered by its own initiating stimulus. This stimulus can be either a signal indicating the completion of the previous link (proprioceptive receptors), an independent external/internal stimulus, or a combination of both. This allows behavior to remain adequate in response to environmental changes or the results of the action itself.
This principle was evolutionarily optimized, starting from the most ancient reaction chains (instincts), whose links are genetically predetermined (unconditioned reflexes). In organisms lacking a system of perceptual primitives formed during ontogenesis (i.e., no neocortex), initiating stimuli are activated by an ancient perceptual system based on subcortical and brainstem structures. This system is capable only of primitive sensory integration and modulation of responses based on the organism’s internal state (hunger, stress, circadian rhythms, etc.), without accounting for environmental specifics the way the neocortex does (whose hierarchy of perceptual primitives reflects the current environment).
Below are three examples of instinctive behavioral chains illustrating the described types of branching:
1. Triggering by a signal indicating completion of the previous link (proprioceptive control)
Instinct: parental care in rodents (e.g., mice)
Initiating stimulus: birth of offspring → activation of maternal behavior.
Link 1: gathering pups into the nest (triggered by olfactory signals from newborns).
Completion of Link 1: all pups are in the nest → proprioceptive signal (e.g., reduced motor activity and tactile excitation).
Link 2: pup care (licking, nursing) is triggered not by a new external stimulus, but by the fact that the previous phase is complete—the system “understands” that gathering is finished and proceeds to the next stage.
Here, the transition is driven by an internal signal of action completion, not by environmental change.
2. Triggering by a new initiating stimulus reflecting a change in circumstances
Instinct: defensive behavior in birds (e.g., a hen encountering a predator)
Initiating stimulus: visual detection of a predator (e.g., a moving shadow overhead).
Link 1: emits an alarm call → chicks freeze or run for cover.
Change in situation: predator approaches → new initiating stimulus (increased angular size of the object, sound of wings).
Link 2: hen either attacks or performs a distraction display (feigning injury)—triggered by the new stimulus, not by completion of the prior action.
The transition is thus driven by updated sensory input signaling threat escalation.
3. Triggering by a combination of a proprioceptive signal and environmental change
Instinct: nest-building in songbirds (e.g., chaffinch)
Initiating stimulus: increased daylight duration + hormonal state → exploratory behavior to select a nest site.
Link 1: site selection and commencement of construction (collecting twigs).
Proprioceptive signal: sufficient depth of the cup-shaped structure has been reached (assessed tactilely/kinesthetically).
Environmental change: rain begins → changes in humidity and temperature.
Link 2: bird switches to collecting waterproof material (moss, feathers)—but only if (a) the nest foundation is already built (proprioceptive signal of phase completion) AND (b) deterioration in weather is detected (new stimulus).
Thus, transitioning to the next link requires both an internal condition (phase completed) and an external one (environmental change).
This demonstrates how the combination of just two fundamental activation criteria—(1) an initiating stimulus (external or internal) signaling a change in conditions, and (2) a proprioceptive signal confirming completion of the prior link—enables evolutionarily ancient, genetically fixed programs (instincts) to achieve high adaptive complexity despite their “rigid” nature.
These two control channels—the event-driven (“what is happening around or inside me?”) and the sequential (“what has already been done?”)—form the foundational architecture for:
Thus, even without a flexible ontogenetically formed perceptual system (like the neocortex), an organism can execute conditionally adequate, multi-level behavior where each action accounts for both the current environmental state and the progress of the behavioral program itself.
Signals indicating changed conditions can arise from two types of sensory changes (internal and external stimuli):
Both types of stimuli, in ancient (pre-neocortical) systems, are handled via innate responses.
The use of these two criteria (stimulus and proprioceptive signal) is the key principle of evolutionary economy: behavioral complexity arises not from arbitrary flexibility, but from a finite yet highly effective set of rules for switching between pre-defined behavioral links.
This principle remains unchanged across all response structures—even up to actions generated by conscious volition. The difference between instinctive and conscious behavior lies not in the fundamental chain structure, but in the origin and flexibility of its components.
At the highest level, an initiating stimulus may be internally generated (e.g., the decision to start writing a text), yet it still serves the same function—triggering a link.
Similarly, the completion signal becomes internalized: instead of proprioceptors, attention and self-observation detect achievement of an intermediate goal (“paragraph written” → ready to proceed to the next).
This means conscious, voluntary behavior is not a rupture, but an evolutionary continuation of the same architecture—where ancient mechanisms are “offloaded” into cortical structures and become manageable via internal models, linguistic constructs, and symbolic planning.
Consequently, the principle of dual control (stimulus + completion state) is truly an invariant across all behavioral levels—from unconditioned reflexes to purposeful creativity.
In general terms, the two invariant criteria are:
(1) a stimulus from a change in response conditions, and
(2) permission to launch the next link, granted by the previous one.
The entire system of branching instinctual chains is built as a rigid logic of hereditary detectors for key environmental changes, which governs chain branching in response to those changes. Under stable conditions, chains unfold solely based on sequential execution criteria.
Overall, the process appears simply as triggering a specific hereditary reaction link in response to a particular condition state—thereby interrupting the active chain and activating a link in a new sequence.
In the context of artificial implementations of living-like agents, this evolutionarily refined universal principle must be used directly, without attempts to circumvent these foundational organizational criteria.
Ignoring this evolutionarily polished behavioral architecture leads to fundamental limitations: fragility, non-adaptiveness, lack of stable motivation, and inability to autonomously set goals in a changing environment. Attempts to replace internal completion signals with external timers, heuristics, or purely stimulus-driven triggers break the causal-goal coherence of behavior. The system loses the ability to “know” what it has already done—and either gets stuck in loops or terminates actions prematurely.
Neglecting motivational context turns the agent into a purely reactive machine, devoid of prioritization. In the real world, this means an inability to choose between “eat” and “flee”—which inevitably leads to failure or collapse.
Therefore, any robustly autonomous artificial system aspiring to adaptive behavior must reproduce this basic architecture—not as a metaphor, but as a functional foundation.
Integration of the neocortex
The emergence of a progressively complex system of perceptual primitive detectors enabled a powerful and flexible pathway for developing new adaptive mechanisms based on prior ones.
When activation of perceptual image detectors in the neocortex coincides with instinct activation—and a mature neuron is present in the motor area of the neocortex—the ancient instinctual chain becomes linked to an abstract image of the action element. This creates an array of detectors capable of triggering the chain within regions controlled by conscious perception systems in the prefrontal cortex. Primarily, this enables empathic activation of action images observed in another subject’s behavior, allowing them to be used as authoritatively mirrored actions for conditions previously not engaged in the observer’s own responses.
When a subject observes another’s action, its sensory systems activate analogous perceptual images, and mirror (or functionally similar) neural systems activate corresponding motor representations.
If, at that moment, the observer’s motivational context is compatible (e.g., also hungry or facing a threat), then the observed action image is not merely recognized—it becomes associated with the observer’s own instinctual program. Given sufficient plasticity, it becomes a new trigger for similar behavior, even in unfamiliar situations.
This is empathic learning—requiring no trial and error, but relying instead on authoritative transfer (“if he acts this way and survives, then it works”).
Maturation of prefrontal structures enables:
Thus, the detector array formed through integration of perception and instincts becomes accessible to conscious control. The subject can mentally “run through” actions, choose among alternatives, and imitate others—not because it “understands” the logic, but because its internal programs are already linked to the corresponding images.
This explains why even the most complex human behaviors—from rituals to scientific creativity—still rely on the same two basic criteria:
(1) a stimulus (perceived or imagined change in conditions), and/or
(2) an internal signal granting permission based on action state.
Any artificial system claiming genuine adaptiveness must not merely simulate intelligence, but reconstruct this hierarchy of embedded motivations, perceptual anchors, and motor linkages—with the capacity for ontogenetic expansion via social interaction. It must possess an innate set of reactions that, during the growth of a hierarchical perceptual recognition tree (fornit.ru/66797), become the foundation for perceptual-motor images of elementary actions. These images are then combined into reaction sequences reflecting instincts, for use within the system of conscious perception-action.
Originally, instincts had no knowledge of condition trees (stimuli). But from the very start of ontogenesis, they become translated into perceptual and action images (fornit.ru/70785), which can then be utilized by conscious volition. Instincts are not replaced—they acquire a perceptual-motor “shell,” enabling them to operate in the space of images, not just reflex arcs.
As a result, newly formed or corrected (now abstract) response chains retain the same continuous branching structure based on the two criteria: stimulus and permission to launch—because this structure was maximally standardized already at the instinctual level.
This architecture is inherently recursive:
Evolutionary extension of the condition tree at the level of conscious responding
A new evolutionary innovation for maintaining response-context integrity is the theme of conscious perception (fornit.ru/71308), organized at the level of a global informational picture that serves as the common context for iterative approximations toward goal-directed actions—particularly novel alternatives to habitual responses under new conditions.
Information about the theme evolves historically: from the simplest signal of the organism’s current state (norm → bad → good), establishing the goal of improving the state when homeostatic parameters deviate from norm; to combinations of basic behavioral styles (feeding, exploratory, sexual, defensive, etc.) forming an abstract emotional image (emotional context); and further to combinations of perceptual conditions characteristic of a conscious awareness theme.
The theme image becomes an extension of the condition tree, acting as a higher-order stimulus that ensures coherence in forming reaction chains during sense-making. This additional branch node becomes part of a new level of historical memory (beyond semantic memory)—namely, episodes of sense-making, enabling storage of response rules based on the significance of conditions and themes, along with evaluation of the reaction’s outcome.
In the artificial prototype of the individual adaptivity system Beast, the condition tree was augmented with “situation” nodes that provide context for the awareness process. While initial automatisms were tied to images from the older condition tree, new automatisms—products of sense-making—began using situation images. In this architecture, it appears feasible to use the theme of awareness itself as the situation node, defining this branch element not artificially (since there’s no clear origin for such a situation array), but as a consciously extracted thematic category. This would greatly unify the architecture—though artificial implementation does not strictly require the situation image to be hereditary. However, without such a foundation, the structure becomes a dead end.
Awareness themes become the highest-order initiating stimulus for forming new reaction chains.
Even in AI, one cannot dispense with an “innate” basis for situations—not in the sense of preloaded scenarios, but in the sense of:
Without this, the condition tree loses semantic depth: it becomes an associative network without a motivational axis—where any feature combination could be labeled a “situation,” yet nothing determines what is truly relevant for survival or adaptation.
This is precisely why, without an innate thematisation function, a system cannot autonomously construct a hierarchy of meaningful contexts.