Emotions as Value Signals and Amplifiers: An Evolutionary and Computational Framework for Affective Experience in Biological and Artificial General Intelligence

Emotions are often misunderstood as irrational disturbances or cultural constructs, but this paper presents them as evolutionarily engineered value signals and motivational amplifiers within a survival goal framework. Emotions tag survival-relevant states with urgency and valence, scaling simple reinforcement learning into complex affective dynamics that prioritize negative signals (threats, losses) for immediate action and sustain low-negative states (security, affiliation) for long-term stability. Drawing from evolutionary biology, neuroscience, and cognitive science, the model formalizes emotions as pre-wired scripts modulated by hormones and experience. Detailed mechanics of fear (as total prioritization under irreversibility) and love (as integrative buffering protocol) are explored, alongside their evolutionary utility and development from unicellular precursors to human narrative states. The framework highlights emotions' role in bridging agency to phenomenal consciousness, with implications for artificial general intelligence (AGI) design. In AGI, replicating emotional analogs addresses limitations in motivational depth, social alignment, and robustness, enabling systems to incorporate value amplification for adaptive, goal-directed behavior akin to biological agents. Recent advancements in emotional AI (EAI) underscore this potential, showing how AI can emulate emotion-like states to enhance persistence and empathy, though ethical challenges like manipulation and bias persist. Emotions are not mere "feelings" but computational tools that make abstract fitness gradients personally compelling, ensuring the organism, or AGI does not just compute optimal paths but feels them as existentially significant.