Статья Principles of Emotion Formation in Artificial Living Creatures
Короткий адрес страницы: fornit.ru/71551 
Озвучка:

Относится к сборнику статей теори МВАП https://t.me/thinking_cycles

Principles of Emotion Formation in Artificial Living Creatures

Russian version
Selection of the Most Rational Method for Constructing Basic Combinations of Behavioral Styles (Emotions)

The homeostatic regulation system in living organisms is based on vital parameters that deviate from their normal range and require corrective actions to restore balance.

 

For each specific deviation scenario, evolutionarily shaped motivational systems activate distinct behavioral styles (e.g., feeding, exploratory, sexual, defensive, etc.) that guide behavior toward restoring homeostasis.

 

For each such behavioral style (hereafter simply “styles”) and their combinations, evolutionary processes have shaped dedicated reaction sequences (instincts).

 

To implement this in an artificial system, the following homeostatic regulation mechanisms can be identified, ensuring an algorithmic sequence:

 
  1. Detection of the degree of deviation from the norm for each vital parameter,
  2. Organization of competition among behavioral styles, taking into account their mutual compatibility (e.g., Anger cannot co-activate with Kindness),
  3. Simultaneous activation of all non-contradictory styles—forming a contextual framework for perception and action aimed at restoring homeostatic balance.
 

In such a system, the active styles serve as recognizers of the current internal state and, in response to external stimuli, trigger specific reactions and reaction chains (see fornit.ru/71450).

 

Before the evolutionary emergence of the neocortex—and the consequent development of hierarchical perception/action representations (fornit.ru/70785)—individual active styles did not form coherent groups. Only with the neocortex could such groups become branches within the perception tree (fornit.ru/66797). Within each such branch, all subsequent perceptual representations become dependent on the emotional state formed by the group of active styles (fornit.ru/70312).

 

At this phylogenetically ancient level (pre-neocortex), behavioral responses are organized as chaotic overlays of reaction chains (“evolutionary patches”). However, with the emergence of a unique ID for each style combination group (i.e., an emotion—represented as an index in biological implementations), it became possible to store the entire hierarchy of representations—from the emotion itself down to the final perceptual episode in memory. This enabled flexible manipulation during problem-solving processes aimed at restoring homeostatic balance.

 

It turns out that storing only the ID of the final perceptual node in a branch is sufficient to reconstruct all intermediate representations—including the emotion itself.

 

In designing an artificial living creature, it becomes essential to define combinations of active styles that:

 

Below is the table of behavioral styles implemented in the artificial creature Beast:

 
 
ID
Basic Context
Purpose of Context
Significance Weight (%)
1
Feeding
Food-seeking behavior, energy replenishment; time-consuming, suppresses antagonistic styles
40
2
Search
Exploratory behavior, curiosity; examining objects of attention, seeking new opportunities
20
3
Play
Play behavior—practicing skills in simplified or educational scenarios
20
4
Mating
Sexual behavior; suppresses antagonistic styles
10
5
Defense
Defensive behavior triggered by clear threat signals or poor internal state
30
6
Laziness
Apathy in states of well-being or hopelessness
5
7
Stupor
Immobility under overwhelming danger or absence of motivation (in well-being or total lack of options)
5
8
Fear
Caution in response to signs of danger
7
9
Aggression
Aggressive behavior toward easy prey or in defense (sometimes in poor internal state)
10
10
Anger
Ruthlessness in response to low evaluation
5
11
Kindness
Altruistic behavior
10
12
Sleep
Sleep state; stress relief and reconstruction of unprocessed information
20

The greater the number of distinct emotions, the more precisely the system can respond to environmental conditions, enabling more diverse and flexible behavioral solutions. The optimal design for any artificial living creature balances minimal sufficiency with maximum differentiation—or, equivalently, maximizing the number of emotions while minimizing the number of styles per emotion and ensuring maximal functional uniqueness.

 

This implies that it is entirely feasible to predefine a fixed number of emotions (Ne) corresponding to the most common life situations and fix the maximum number of styles per emotion (Ns). The creature will then develop on this basis, with its behavioral repertoire constrained primarily by Ne, and only weakly dependent on Ns—provided Ns ≥ 2 (since the most significant styles dominate, and including 2–3 is usually sufficient; additional styles have minimal impact).

 

Crucially, the number of emotions (Ne)—not the number of base styles (Ns)—determines the expressive and adaptive power of the system. Therefore, it is reasonable to fix Ns = 3 (maximum of three styles per emotion).

 

For Ne = 12 base styles and Ns = 3, the total number of non-redundant style combinations (ignoring antagonisms and treating combinations as unordered sets) equals 1,680,592.

 

However, assigning innate reaction patterns and instincts to each of these millions of hypothetical emotions would be computationally and temporally infeasible—and practically unnecessary. The gain in behavioral flexibility would be marginal compared to a system with just ~1,000 well-chosen emotions.

 

Indeed, the vast majority of these combinations would describe negligible situational nuances that do not require unique behavioral responses.

 

Moreover, in urgent survival situations (e.g., suffocation), motivational competition becomes so sharply prioritized that combinatorial methods ignoring this hierarchy are invalid. For instance, the need to breathe overrides all other motivations—only behaviors aimed at restoring respiration are activated. Similarly, thirst, extreme temperature regulation, and hunger follow a strict priority order.

 

Thus, under extreme deviation from homeostasis, Ns should be limited to 1 (the highest-priority style only). Likewise, in the “Good” state—when a need is already being satisfied and the organism is maintaining that state—Ns should also be 1. Only in the “Normal” state (or minor deviations) is it safe and beneficial to combine multiple styles.

 

Ns = 1 ensures:

 

The “Normal” state (or minor deviations) allows the system to:

 

In the Beast artificial system, Ns was never limited in “Bad” or “Good” states (though always ≤3), but during heuristic design of innate reactions, the excess of style combinations felt clearly redundant.

 

 

Scientific Background

Several key theories and research programs support the view that emotions emerge from more fundamental motivational systems—though often using different terminology: motivational systems, behavioral modes, action systems, primary processes, survival circuits, etc.

 
 

Both Anokhin’s and Ukhtomsky’s frameworks treat emotions as derivatives of the current motivational state, which itself arises from parameter deviations.

 

 

Empirical Estimates of Human Emotion Discrimination

Empirical studies estimate that humans can reliably distinguish 20–30 basic and mixed emotions, and with fine-grained analysis, up to several hundred emotional states.

 
 

Neuroimaging and behavioral data suggest the brain encodes emotions not as discrete categories, but as points in a continuous multidimensional space (e.g., valence, arousal, control, novelty).

 

However, in verbalization and recognition, humans use a limited set of stable labels—typically 30–50 in everyday life.

 

Linguistically, while hundreds of emotion terms exist across languages, most are contextual nuances or blends (e.g., “tоска” [Russian melancholy], “hope”, “shame”). The core emotional lexicon for native speakers is typically 20–30 concepts.

 

Thus, empirically, the functionally significant and discriminable number of human emotions is 20–30, extendable to ~50–100 with cultural and individual nuances.

 

This strongly supports designing emotion systems by selecting style combinations based on realistic life situations, with Ns adjusted according to global state: Bad, Good, or Normal (fornit.ru/70332).

 

 

Proposed Emotions for Artificial Creatures

Base Behavioral Styles:

  1. Hunger (Foraging)
  2. Search (Curiosity / Exploration)
  3. Play (Training / Recreation)
  4. Comfort (Sleep / Warmth / Cleanliness)
  5. Defense (Active aggression / attack)
  6. Escape (Passive threat avoidance)
  7. Reproduction (Sexual behavior)
  8. Care (Maternal / social instinct)
  9. Territoriality (Marking, zone defense)
  10. Sociality (Attachment, grooming)
  11. Cooling (Avoiding overheating)
  12. Energy Conservation (Minimizing effort)
 

 

Emotions for “BAD” State (Cat) — Ns = 1

(Critical deviation; single-style focus)

 
 
ID
Style(s)
Purpose
P1
S1
Urgent hunger satisfaction
P2
S5
Immediate self-defense against attack
P3
S6
Rapid escape from danger
P4
S11
Urgent cooling (overheating)
P5
S9
Aggressive territorial defense
P6
S7
Urgent mating behavior (estrus/hunting)

Emotions for “GOOD” State (Cat) — Ns = 1

(Need satisfied; maintaining favorable state)

 
 
ID
Style(s)
Purpose
G1
S4
Enjoying comfort (sleeping, warmth)
G2
S12
Conscious rest, effort minimization
G3
S10
Affection display (purring, rubbing)
G4
S8
Caring for kittens (licking, feeding)
G5
S7
Sexual satisfaction
G6
S3
Play as an end in itself (process enjoyment)

Emotions for “NORMAL” State (Cat) — Ns = 2–3

(Baseline mode; complex behavior via style combinations)

 
 
ID
Style(s)
Purpose
N1
S2, S3
Exploratory play with novel objects
N2
S2, S9
Territory patrolling and marking
N3
S1, S2
Food search without acute hunger
N4
S10, S3
Social play with conspecific or human
N5
S4, S12
Relaxed rest in safe environment
N6
S7, S2
Mate searching
N7
S8, S10
Social grooming, caring for "in-group"
N8
S5, S9
Territory boundary patrol and defense
N9
S2, S11
Seeking cool spots on warm days
N10
S1, S9, S5
Aggressive defense of food resources on territory
N11
S6, S12
Energy-efficient threat avoidance
N12
S3, S10, S12
Calm social play without exhaustion

 

Emotions for “BAD” State (Human) — Ns = 1

 
ID
Style(s)
Purpose
P1
H1
Acute physiological need (hunger, pain)
P2
H5
Immediate self-defense, rage
P3
H6
Panic, escape from danger
P4
H4
Severe anxiety about safety
P5
H9
Acute humiliation or status loss
P6
H10
Intense loneliness or rejection

Emotions for “GOOD” State (Human) — Ns = 1

 
ID
Style(s)
Purpose
G1
H1
Satiety, thirst quenching
G2
H4
Sense of safety, comfort
G3
H10
Love, belonging
G4
H9
Respect, recognition
G5
H7
Sexual satisfaction
G6
H3
Flow state, creative uplift

Emotions for “NORMAL” State (Human) — Ns = 2–3

 
ID
Style(s)
Purpose
N1
H2, H3
Scientific inquiry, creative exploration
N2
H2, H11
Goal-directed learning, skill acquisition
N3
H10, H8
Care in close relationships
N4
H9, H11
Career advancement, achievement drive
N5
H7, H10
Romantic involvement, flirting
N6
H3, H10
Collaborative creativity, social play
N7
H4, H12
Mindful rest, relaxation
N8
H5, H9
Competition, defending interests
N9
H8, H11
Mentorship
N10
H2, H4, H12
Comfortable learning without stress
N11
H9, H10, H8
Social leadership, group responsibility
N12
H3, H7, H10
Playful flirting, light social interaction
N13
H5, H6, H4
Cautious risk assessment, vigilance
N14
H1, H2, H3
Seeking novel culinary experiences

As seen, emotional architecture need not differ fundamentally between cats and humans. However, for creative or unsatisfied individuals, the set of “Normal”-state emotions can be expanded:

 

(Extended list of 30 human “Normal”-state emotions, N15–N30, follows with combinations like “Exploring personal limits”, “Public creativity”, “Family comfort creation”, “Strategic conflict avoidance”, etc.—all built from meaningful, adaptive behavioral style combinations.)

 

Notably, this heuristic approach avoids the need for explicit antagonist tables or style-weight calculations, as combinations are selected based on adaptive behavioral meaning. This intuitive, context-driven selection proves far more efficient and effective than brute-force combinatorial methods or blind generalization.

 

In conclusion, overly meticulous automated methods for emotion selection yield diminishing returns compared to heuristic, functionally grounded design.


Авторизованные пользователи могут оставлять комментарии.