MAYA · THE GROWING MIND
Age 21
Full Antahkarana · Prana complete · Series closed
1.000
PRANA
Buddhi
1.000
Vairagya
46%
Karma
0.057
Pruned
84%
Bhaya
0.000
NAVIGATE
“Without Prana, nothing moves, nothing learns, nothing grows. Every other dimension of the Antahkarana operates within the budget Prana provides. Across nine papers, we have demonstrated the computational maturation of a mind.”
Maya-Prana · Paper 9 · Nexus Learning Labs · Series Conclusion
§1 WHAT IS PRANA?
Plain language: no computer science needed
Imagine a student studying for exams. The harder they study, the more mentally exhausted they become. If they study without rest, their ability to absorb new information degrades. Sleep restores their capacity. This is Prana — the metabolic budget that governs how much the brain can change per unit time.
In biological brains, this is governed by the Astrocyte-Neuron Lactate Shuttle (ANLS). Astrocytes supply lactate (fuel) to active neurons. When learning demand exceeds supply, plasticity degrades. Maya-Prana models this as a scalar budget that depletes under gradient load and recovers during rest — exactly as the biology prescribes.
prana -= PRANA_COST_RATE × gradient_magnitude × activity_level
if activity_level < threshold: prana += RECOVERY_RATE × (1 − activity) × (0.5 + vairagya × 0.5)
effective_lr = base_lr × prana × (0.5 + buddhi × 0.5)
Prana is not a weight mask. Not a gradient gate. It is the metabolic budget — the fuel available for plasticity.
§2 THE MAYA SERIES · P1 TO P9
A 9-paper arc: from a pain signal to a complete mind
P1 Bhaya
P2 OS
P3 Vairagya
P4 Buddhi
P5 Viveka
P6 Chitta
P7 Manas
P8 Karma
P9 Prana ✦
Click any node above to learn about that paper.
P1–P7: Maya age 5 — reactive, pain-driven, Buddhi nascent. P8: First paper of maturation. P9: Maya age 21 — full Antahkarana, Prana complete.
§3 ALL 9 ANTAHKARANA DIMENSIONS
Each dimension is a falsifiable computational mechanism — not metaphor
भय
BHAYA
Fear
Pain signal triggers hyper-plasticity. Offensive metaplasticity — the opposite of TACOS.
P1 · DOI
वैराग्य
VAIRAGYA
Detachment
Earned protection from interference. Synapses that survive pain accumulate Vairagya and resist future overwriting.
P1/P3 · DOI
बुद्धि
BUDDHI
Intellect
Discriminative S-curve gate. Accumulates with experience. Young Maya uses energy impulsively; mature Maya uses it wisely.
P4 · DOI
विवेक
VIVEKA
Discernment
Cross-task consistency tracking. Identifies synapses that fire reliably across tasks and amplifies their Vairagya protection.
P5 · DOI
चित्त
CHITTA
Subconscious store
Retrograde gradient gate. Synapses with high Samskara suppress their own gradients — implicit memory slows overwriting.
P6 · DOI
मनस्
MANAS
The doubting mind
Oscillatory LIF threshold. Descends across timesteps — from suppression to receptivity. Only salient signals survive the high threshold.
P7 · DOI
कर्म
KARMA
Consequence
Second-order plasticity history. Absolute integral of weight trajectory changes across tasks. High Karma = chronic cross-task interference.
P8 · DOI
शून्यता
SHUNYATA
Emptiness / Release
Structured synaptic pruning at task boundaries. Biologically grounded in microglial phagocytosis via C1q/C3 complement cascade.
P8 · DOI
प्राण
PRANA
Vital life force
Metabolic plasticity budget. Gates the learning rate per batch. Depletes under gradient load, recovers at rest. Modelled on the Astrocyte-Neuron Lactate Shuttle (ANLS). Never depleted at canonical cost rate — biologically accurate.
P9 ✦ SERIES COMPLETE
§4 LIVE HYPERPARAMETERS
Drag sliders to see how each parameter affects the mechanism. Press R to reset.
PRANA_COST_RATE0.002315
BUDDHI (simulated)0.982
PRANA_RECOVERY_RATE0.05
KARMA_THRESHOLD0.05
PRANA_MIN (floor)0.05
VAIRAGYA_PROTECTION0.30
effective_lr = base_lr × prana × (0.5 + buddhi × 0.5)
effective_lr = 0.009910
effective_threshold = karma_base × (0.5 + buddhi × 0.5)
effective_threshold = 0.0495
§5 ABLATION CONDITIONS · 6 EXPERIMENTS
Each condition isolates one variable. Click to explore.
Condition A
Baseline
12.46%
BWT −53.80%
Condition B
Prana only
10.33%
BWT −55.82%
Condition C
Fixed Prana
12.02%
BWT −54.82%
Condition D ★
Canonical
12.72%
BWT −54.32%
Condition E
Aggressive
12.25%
BWT −54.70%
Condition F ◆
No Buddhi gate
13.68%
BWT −51.20% · Best
Condition D (canonical ★): Full Maya-Prana stack. Prana gates effective_lr per batch. Buddhi modulates the gate. Karma accumulates interference; Shunyata prunes at boundaries. All 9 dimensions active simultaneously.
§6 RESULTS · AFFECTIVE DYNAMICS
Prana trajectory across tasks (canonical)
Effective LR across tasks — D vs F
Buddhi S-curve (series constant P4–P9)
Cumulative pruning by task boundary
ACCURACY MATRIX · CONDITION D (CANONICAL ★)
| After T→ | T0 | T1 | T2 | T3 | T4 | T5 | T6 | T7 | T8 | T9 |
|---|
§7 SERIES CONSTANTS · 9 CONFIRMATIONS
Bhaya Quiescence Law · 9th consecutive confirmation
Bhaya = 0.000 under replay · All conditions · All 9 papers
Bhaya is driven by replay — use as catastrophic forgetting monitor.
Buddhi S-curve · Architecturally deterministic
Identical trajectory P4–P9 · All ablation conditions
Structural property of the experience accumulation formula.
BHAYA QUIESCENCE HEATMAP · ALL 9 PAPERS
Green = 0.000 (quiescent). Every cell in every paper is green under replay.
§8 KEY FINDINGS
FINDING 01
Prana Resilience
At PRANA_COST_RATE=0.002315, Prana maintained full budget (1.0000) across all 10 tasks. Even at 3.5x the cost rate, Prana never depleted. Biologically accurate — the ANLS does not fail under standard cognitive load.
FINDING 02 · UNEXPECTED
Condition F: The Surprise
Removing Buddhi modulation (fixed EffLR=0.0075) produced the best result: AA=13.68%, BWT=-51.2%, Pruned=46.93%. The Buddhi warm-up penalises early task consolidation. A consistent learning rate outperformed the maturity-dependent schedule.
FINDING 03
C ≈ D — Fixed Prana Matches Canonical
Condition C (fixed Prana=1.0) produced AA=12.02% vs D=12.72%. Near-equivalence confirms Prana depletion had minimal effect at the canonical cost rate — Prana is resilient, not absent.
FINDING 04
B Degrades Without Full Antahkarana
Condition B (Prana only) produced the worst result: AA=10.33%, Pruned=91.85%. Prana alone cannot compensate for the loss of other affective dimensions. The Antahkarana is an integrated system.
§9 LIVE EMBODIED DEMONSTRATION
Maya running in a physical robot — PiCar-X with Raspberry Pi 5, full Antahkarana active
Live Demo can be watched here
PiCar-X embodied Maya · Full Antahkarana · All 9 dimensions live · Coming soon
PAPER · CODE · RESEARCH JOURNEY
“Across nine papers, we have demonstrated the computational maturation of a mind.”