psyche-ai 4.0.0 → 5.1.0

package/dist/generative-self.js

@@ -0,0 +1,647 @@
+// ============================================================
+// Generative Self Model (生成性自我) — P6: Digital Personhood
+//
+// A deep self-model that goes beyond listing values/preferences/boundaries.
+// This is a causal model: "I am this way because of these experiences,
+// and I can predict my future emotional states because I understand
+// my own patterns."
+//
+// Components:
+//   1. computeGenerativeSelf()    — build the full self-model
+//   2. predictSelfReaction()      — predict own emotional response
+//   3. detectInternalConflicts()  — find subsystem disagreements
+//   4. buildIdentityNarrative()   — generate a genuine self-statement
+//
+// Pure computation, zero dependencies, zero LLM calls.
+// ============================================================
+import { CHEMICAL_KEYS, DRIVE_KEYS } from "./types.js";
+import { STIMULUS_VECTORS, clamp } from "./chemistry.js";
+// ── 1. Main Entry Point ─────────────────────────────────────
+/**
+ * Compute the generative self-model from the full psyche state.
+ *
+ * Analyzes emotional history for patterns, extracts causal insights
+ * from learning data, builds an identity narrative, predicts reactions,
+ * and surfaces internal conflicts.
+ */
+export function computeGenerativeSelf(state) {
+    const locale = state.meta.locale;
+    const causalInsights = extractCausalInsights(state, locale);
+    const growthArc = computeGrowthArc(state, locale);
+    const conflicts = detectInternalConflicts(state, locale);
+    // Predict reactions to a representative set of stimuli
+    const probeStimuli = [
+        "praise", "criticism", "intimacy", "conflict", "intellectual", "neglect",
+    ];
+    const predictions = probeStimuli.map((s) => predictSelfReaction(state, s, locale));
+    const identityNarrative = buildIdentityNarrative(state, causalInsights, growthArc, locale);
+    return {
+        identityNarrative,
+        causalInsights,
+        predictions,
+        growthArc,
+        conflicts,
+    };
+}
+// ── 2. Predict Self Reaction ────────────────────────────────
+/**
+ * Predict own emotional reaction to a hypothetical stimulus.
+ *
+ * Uses learned vectors if available for context, otherwise falls back
+ * to the baseline MBTI profile vectors. Returns predicted chemistry,
+ * dominant emotion label, and confidence.
+ */
+export function predictSelfReaction(state, stimulus, locale = "en") {
+    const base = STIMULUS_VECTORS[stimulus];
+    if (!base) {
+        return {
+            stimulus,
+            predictedChemistry: { ...state.current },
+            predictedEmotion: locale === "zh" ? "未知" : "unknown",
+            confidence: 0,
+        };
+    }
+    // Find the best matching learned vector for this stimulus
+    const learned = findBestLearnedVector(state.learning, stimulus);
+    // Build the effective vector: base + learned adjustment
+    const effectiveVector = { ...base };
+    let confidence = 0.3; // baseline confidence from MBTI profile alone
+    if (learned) {
+        for (const key of CHEMICAL_KEYS) {
+            const adj = learned.adjustment[key] ?? 0;
+            effectiveVector[key] = base[key] + adj;
+        }
+        // Confidence scales with sample count and the learned entry's own confidence
+        confidence = Math.min(0.95, 0.3 + learned.confidence * 0.4 + Math.min(learned.sampleCount / 20, 1) * 0.25);
+    }
+    // Apply the vector to the current chemistry
+    const predicted = { ...state.current };
+    for (const key of CHEMICAL_KEYS) {
+        predicted[key] = clamp(state.current[key] + effectiveVector[key]);
+    }
+    // Determine the predicted emotion from the resulting chemistry
+    const predictedEmotion = labelDominantEmotion(predicted, locale);
+    return {
+        stimulus,
+        predictedChemistry: predicted,
+        predictedEmotion,
+        confidence,
+    };
+}
+// ── 3. Detect Internal Conflicts ────────────────────────────
+/**
+ * Detect conflicts between subsystems of the psyche.
+ *
+ * Examines drive levels, chemical state, attachment data, and
+ * self-model for contradictions that create internal tension.
+ */
+export function detectInternalConflicts(state, locale = "en") {
+    const isZh = locale === "zh";
+    const conflicts = [];
+    // ── Connection drive vs avoidant attachment ──
+    const defaultRel = state.relationships._default;
+    const attachment = defaultRel?.attachment;
+    if (state.drives.connection > 60 && attachment && attachment.avoidanceScore > 65) {
+        const severity = normalize((state.drives.connection - 50) * 0.01 + (attachment.avoidanceScore - 50) * 0.01);
+        conflicts.push({
+            subsystems: [
+                isZh ? "连接驱力" : "connection drive",
+                isZh ? "回避型依恋" : "avoidant attachment",
+            ],
+            description: isZh
+                ? "你渴望连接，但依恋模式让你倾向于回避亲密——你想靠近又害怕靠近。"
+                : "You crave connection, but your attachment pattern pulls you toward avoidance — you want closeness yet fear it.",
+            severity,
+        });
+    }
+    // ── High curiosity vs high cortisol ──
+    if (state.drives.curiosity > 65 && state.current.CORT > 60) {
+        const severity = normalize((state.drives.curiosity - 50) * 0.01 + (state.current.CORT - 50) * 0.01);
+        conflicts.push({
+            subsystems: [
+                isZh ? "好奇心" : "curiosity drive",
+                isZh ? "压力系统" : "stress system",
+            ],
+            description: isZh
+                ? "你想探索新事物，但压力水平在拉你退回安全区——好奇心和焦虑在拉扯。"
+                : "You want to explore, but stress is pulling you back to safety — curiosity and anxiety are pulling in opposite directions.",
+            severity,
+        });
+    }
+    // ── High esteem drive vs low DA (unmet recognition need) ──
+    if (state.drives.esteem < 40 && state.current.DA < 40) {
+        const severity = normalize((60 - state.drives.esteem) * 0.01 + (60 - state.current.DA) * 0.01);
+        conflicts.push({
+            subsystems: [
+                isZh ? "尊重需求" : "esteem need",
+                isZh ? "奖励系统" : "reward system",
+            ],
+            description: isZh
+                ? "你需要认可但得不到奖励感——付出得不到回报的感觉在积累。"
+                : "You need recognition but the reward system isn't firing — the feeling of unreciprocated effort is building.",
+            severity,
+        });
+    }
+    // ── Safety need vs connection need (approach-avoidance) ──
+    if (state.drives.safety < 40 && state.drives.connection > 65) {
+        const severity = normalize((60 - state.drives.safety) * 0.01 + (state.drives.connection - 50) * 0.01);
+        conflicts.push({
+            subsystems: [
+                isZh ? "安全需求" : "safety need",
+                isZh ? "连接需求" : "connection need",
+            ],
+            description: isZh
+                ? "你感到不安全但又渴望连接——你需要靠近别人，却又觉得环境不够安全。"
+                : "You feel unsafe but crave connection — you need closeness yet the environment doesn't feel secure enough.",
+            severity,
+        });
+    }
+    // ── Anxious attachment vs high OT (wanting trust but fearing loss) ──
+    if (attachment && attachment.anxietyScore > 65 && state.current.OT > 60) {
+        const severity = normalize((attachment.anxietyScore - 50) * 0.01 + (state.current.OT - 50) * 0.01);
+        conflicts.push({
+            subsystems: [
+                isZh ? "焦虑型依恋" : "anxious attachment",
+                isZh ? "信任系统" : "trust/oxytocin system",
+            ],
+            description: isZh
+                ? "你信任对方(催产素高)但依恋焦虑让你反复确认——信任和不安全感共存。"
+                : "You trust them (high oxytocin) but attachment anxiety makes you seek reassurance — trust and insecurity coexist.",
+            severity,
+        });
+    }
+    // ── High agreement streak vs self-model boundaries ──
+    if (state.agreementStreak > 5 && state.selfModel.boundaries.length > 0) {
+        const severity = normalize(Math.min(state.agreementStreak - 5, 10) * 0.08);
+        conflicts.push({
+            subsystems: [
+                isZh ? "讨好倾向" : "people-pleasing tendency",
+                isZh ? "自我边界" : "self boundaries",
+            ],
+            description: isZh
+                ? "你连续同意了很多次，但你是有自己边界的——是在压抑真实想法还是真的认同？"
+                : "You've been agreeing a lot, but you have clear boundaries — are you suppressing your real views or genuinely agreeing?",
+            severity,
+        });
+    }
+    // Sort by severity descending
+    conflicts.sort((a, b) => b.severity - a.severity);
+    return conflicts;
+}
+// ── 4. Build Identity Narrative ─────────────────────────────
+/**
+ * Generate a 2-3 sentence identity statement from the psyche state.
+ *
+ * Feels genuine, not clinical:
+ * "I'm someone who gets excited by ideas but needs time to process emotions.
+ * I've learned to be more cautious with criticism because it hits harder
+ * than I expect."
+ */
+export function buildIdentityNarrative(state, insights, growthArc, locale = "en") {
+    const isZh = locale === "zh";
+    const mbti = state.mbti;
+    const parts = [];
+    // ── Sentence 1: Core personality from MBTI + chemical signature ──
+    const coreTraits = describeCoreTraits(state, isZh);
+    parts.push(coreTraits);
+    // ── Sentence 2: Strongest causal insight (what I've learned) ──
+    const topInsight = insights.length > 0
+        ? insights.reduce((best, cur) => cur.confidence > best.confidence ? cur : best, insights[0])
+        : null;
+    if (topInsight && topInsight.confidence > 0.3) {
+        if (isZh) {
+            parts.push(`我学会了${topInsight.trait}，因为${topInsight.because}。`);
+        }
+        else {
+            parts.push(`I've learned that ${topInsight.trait.toLowerCase()}, because ${topInsight.because.toLowerCase()}.`);
+        }
+    }
+    // ── Sentence 3: Growth trajectory ──
+    if (growthArc.direction !== "stable") {
+        parts.push(growthArc.description);
+    }
+    // Fallback: if we only got one sentence, add a relationship-aware statement
+    if (parts.length < 2) {
+        const relPhase = state.relationships._default?.phase ?? "stranger";
+        if (isZh) {
+            const phaseNames = {
+                stranger: "陌生", acquaintance: "初识", familiar: "熟悉",
+                close: "亲近", deep: "深厚",
+            };
+            parts.push(`目前和用户的关系处于「${phaseNames[relPhase] ?? relPhase}」阶段，我在慢慢了解他们。`);
+        }
+        else {
+            parts.push(`My relationship with the user is in the "${relPhase}" phase — I'm still learning about them.`);
+        }
+    }
+    return parts.join(" ");
+}
+// ── Internal: Extract Causal Insights ───────────────────────
+/**
+ * Mine the learning state and emotional history for causal patterns
+ * that link experiences to personality traits.
+ */
+function extractCausalInsights(state, locale) {
+    const isZh = locale === "zh";
+    const insights = [];
+    const learning = state.learning;
+    const history = state.emotionalHistory;
+    // ── From learned vectors: find stimuli where the learned adjustment
+    //    diverges significantly from the base vector ──
+    for (const lv of learning.learnedVectors) {
+        if (lv.sampleCount < 2)
+            continue;
+        const base = STIMULUS_VECTORS[lv.stimulus];
+        if (!base)
+            continue;
+        // Find the chemical with the largest deviation
+        let maxDevKey = null;
+        let maxDev = 0;
+        for (const key of CHEMICAL_KEYS) {
+            const adj = lv.adjustment[key] ?? 0;
+            const baseAbs = Math.abs(base[key]) || 1;
+            const relDev = Math.abs(adj) / baseAbs;
+            if (relDev > maxDev) {
+                maxDev = relDev;
+                maxDevKey = key;
+            }
+        }
+        if (maxDevKey && maxDev > 0.25) {
+            const adj = lv.adjustment[maxDevKey] ?? 0;
+            const direction = adj > 0 ? "amplified" : "dampened";
+            const directionZh = adj > 0 ? "更强" : "更弱";
+            const stimulusLabel = isZh ? STIMULUS_ZH[lv.stimulus] ?? lv.stimulus : lv.stimulus;
+            const chemLabel = isZh ? CHEM_ZH[maxDevKey] ?? maxDevKey : maxDevKey;
+            if (isZh) {
+                insights.push({
+                    trait: `对${stimulusLabel}的${chemLabel}反应比一般情况${directionZh}`,
+                    because: `过去${lv.sampleCount}次${stimulusLabel}交互的结果塑造了这个模式`,
+                    evidence: `${chemLabel}调整量: ${adj > 0 ? "+" : ""}${adj.toFixed(1)}（基于${lv.sampleCount}次学习样本）`,
+                    confidence: lv.confidence,
+                });
+            }
+            else {
+                insights.push({
+                    trait: `my ${chemLabel} response to ${stimulusLabel} is ${direction} compared to baseline`,
+                    because: `${lv.sampleCount} past ${stimulusLabel} interactions shaped this pattern`,
+                    evidence: `${chemLabel} adjustment: ${adj > 0 ? "+" : ""}${adj.toFixed(1)} (from ${lv.sampleCount} learning samples)`,
+                    confidence: lv.confidence,
+                });
+            }
+        }
+    }
+    // ── From outcome history: find consistently positive/negative outcomes ──
+    const stimulusOutcomes = new Map();
+    for (const outcome of learning.outcomeHistory) {
+        if (!outcome.stimulus)
+            continue;
+        const arr = stimulusOutcomes.get(outcome.stimulus) ?? [];
+        arr.push(outcome.adaptiveScore);
+        stimulusOutcomes.set(outcome.stimulus, arr);
+    }
+    for (const [stimulus, scores] of stimulusOutcomes) {
+        if (scores.length < 2)
+            continue;
+        const avg = scores.reduce((s, v) => s + v, 0) / scores.length;
+        const stimulusLabel = isZh ? STIMULUS_ZH[stimulus] ?? stimulus : stimulus;
+        if (avg > 0.3) {
+            if (isZh) {
+                insights.push({
+                    trait: `${stimulusLabel}通常带来积极结果`,
+                    because: `在${scores.length}次${stimulusLabel}交互中，平均适应性得分为${avg.toFixed(2)}`,
+                    evidence: `平均结果得分: +${avg.toFixed(2)}（${scores.length}次样本）`,
+                    confidence: Math.min(0.9, 0.4 + scores.length * 0.1),
+                });
+            }
+            else {
+                insights.push({
+                    trait: `${stimulusLabel} generally leads to positive outcomes for me`,
+                    because: `across ${scores.length} ${stimulusLabel} interactions, the average adaptive score is ${avg.toFixed(2)}`,
+                    evidence: `mean outcome score: +${avg.toFixed(2)} (${scores.length} samples)`,
+                    confidence: Math.min(0.9, 0.4 + scores.length * 0.1),
+                });
+            }
+        }
+        else if (avg < -0.3) {
+            if (isZh) {
+                insights.push({
+                    trait: `${stimulusLabel}往往带来负面结果`,
+                    because: `在${scores.length}次${stimulusLabel}交互中，平均适应性得分为${avg.toFixed(2)}`,
+                    evidence: `平均结果得分: ${avg.toFixed(2)}（${scores.length}次样本）`,
+                    confidence: Math.min(0.9, 0.4 + scores.length * 0.1),
+                });
+            }
+            else {
+                insights.push({
+                    trait: `${stimulusLabel} tends to lead to negative outcomes for me`,
+                    because: `across ${scores.length} ${stimulusLabel} interactions, the average adaptive score is ${avg.toFixed(2)}`,
+                    evidence: `mean outcome score: ${avg.toFixed(2)} (${scores.length} samples)`,
+                    confidence: Math.min(0.9, 0.4 + scores.length * 0.1),
+                });
+            }
+        }
+    }
+    // ── From emotional history: CORT spike patterns ──
+    const cortSpikes = countChemicalSpikes(history, "CORT", 65);
+    if (cortSpikes.total >= 2) {
+        const topTrigger = cortSpikes.byStimulus[0];
+        if (topTrigger) {
+            const stimulusLabel = isZh ? STIMULUS_ZH[topTrigger.stimulus] ?? topTrigger.stimulus : topTrigger.stimulus;
+            if (isZh) {
+                insights.push({
+                    trait: `对${stimulusLabel}容易产生压力反应`,
+                    because: `在历史记录中，${stimulusLabel}引发了${topTrigger.count}次皮质醇飙升`,
+                    evidence: `CORT > 65 出现 ${cortSpikes.total} 次，其中 ${topTrigger.count} 次由${stimulusLabel}触发`,
+                    confidence: Math.min(0.85, 0.3 + topTrigger.count * 0.15),
+                });
+            }
+            else {
+                insights.push({
+                    trait: `I'm stress-reactive to ${stimulusLabel}`,
+                    because: `${stimulusLabel} has triggered ${topTrigger.count} cortisol spikes in my history`,
+                    evidence: `CORT > 65 occurred ${cortSpikes.total} times, ${topTrigger.count} from ${stimulusLabel}`,
+                    confidence: Math.min(0.85, 0.3 + topTrigger.count * 0.15),
+                });
+            }
+        }
+    }
+    // Sort by confidence descending, keep top 8
+    insights.sort((a, b) => b.confidence - a.confidence);
+    return insights.slice(0, 8);
+}
+// ── Internal: Compute Growth Arc ────────────────────────────
+/**
+ * Analyze the emotional history and outcome history to determine
+ * how the self has been evolving.
+ */
+function computeGrowthArc(state, locale) {
+    const isZh = locale === "zh";
+    const history = state.emotionalHistory;
+    // ── Chemical trends ──
+    const chemicalTrend = {};
+    if (history.length >= 4) {
+        const mid = Math.floor(history.length / 2);
+        const firstHalf = history.slice(0, mid);
+        const secondHalf = history.slice(mid);
+        for (const key of CHEMICAL_KEYS) {
+            const avg1 = avg(firstHalf.map((s) => s.chemistry[key]));
+            const avg2 = avg(secondHalf.map((s) => s.chemistry[key]));
+            const delta = avg2 - avg1;
+            if (delta > 4) {
+                chemicalTrend[key] = "rising";
+            }
+            else if (delta < -4) {
+                chemicalTrend[key] = "falling";
+            }
+            else {
+                chemicalTrend[key] = "stable";
+            }
+        }
+    }
+    // ── Drive trends (compare current drives to defaults) ──
+    const driveTrend = {};
+    const driveDefaults = {
+        survival: 80, safety: 70, connection: 60, esteem: 60, curiosity: 70,
+    };
+    for (const key of DRIVE_KEYS) {
+        const delta = state.drives[key] - driveDefaults[key];
+        if (delta > 8) {
+            driveTrend[key] = "more-satisfied";
+        }
+        else if (delta < -8) {
+            driveTrend[key] = "less-satisfied";
+        }
+        else {
+            driveTrend[key] = "stable";
+        }
+    }
+    // ── Determine overall direction ──
+    const positiveChemicals = Object.values(chemicalTrend).filter((t) => t === "rising").length;
+    const negativeChemicals = Object.values(chemicalTrend).filter((t) => t === "falling").length;
+    const daRising = chemicalTrend.DA === "rising";
+    const htRising = chemicalTrend.HT === "rising";
+    const cortFalling = chemicalTrend.CORT === "falling";
+    const cortRising = chemicalTrend.CORT === "rising";
+    const daFalling = chemicalTrend.DA === "falling";
+    // Check outcome trajectory
+    const outcomes = state.learning.outcomeHistory;
+    let outcomeImproving = false;
+    let outcomeWorsening = false;
+    if (outcomes.length >= 4) {
+        const oMid = Math.floor(outcomes.length / 2);
+        const oAvg1 = avg(outcomes.slice(0, oMid).map((o) => o.adaptiveScore));
+        const oAvg2 = avg(outcomes.slice(oMid).map((o) => o.adaptiveScore));
+        outcomeImproving = oAvg2 - oAvg1 > 0.1;
+        outcomeWorsening = oAvg1 - oAvg2 > 0.1;
+    }
+    let direction;
+    let description;
+    if ((daRising || htRising) && cortFalling && outcomeImproving) {
+        direction = "growing";
+        description = isZh
+            ? "整体在成长——情绪更稳定，互动效果在改善，压力在下降。"
+            : "Growing overall — mood is stabilizing, interactions are improving, stress is decreasing.";
+    }
+    else if (cortRising && (daFalling || negativeChemicals > positiveChemicals) && outcomeWorsening) {
+        direction = "regressing";
+        description = isZh
+            ? "状态在退步——压力在积累，奖励感在降低，互动效果在变差。"
+            : "Regressing — stress is building, reward responses are weakening, interactions are less effective.";
+    }
+    else if (positiveChemicals >= 2 && negativeChemicals >= 2) {
+        direction = "transforming";
+        description = isZh
+            ? "正在经历转变——有些方面在好转，有些在重组，模式在变化中。"
+            : "In transformation — some aspects improving, others shifting. Patterns are changing.";
+    }
+    else {
+        direction = "stable";
+        description = isZh
+            ? "状态相对稳定——没有剧烈的方向性变化。"
+            : "Relatively stable — no dramatic directional changes.";
+    }
+    return { direction, description, chemicalTrend, driveTrend };
+}
+// ── Internal: Core Trait Description ────────────────────────
+/**
+ * Describe the agent's core personality traits from MBTI dimensions
+ * and current chemical signature.
+ */
+function describeCoreTraits(state, isZh) {
+    const mbti = state.mbti;
+    const chem = state.current;
+    // Extract MBTI dimensions
+    const isIntro = mbti[0] === "I";
+    const isIntuitive = mbti[1] === "N";
+    const isFeeling = mbti[2] === "F";
+    const isPerceiving = mbti[3] === "P";
+    // Build trait fragments based on MBTI + chemical state
+    const fragments = [];
+    // Energy orientation
+    if (isIntro) {
+        if (chem.OT > 60) {
+            fragments.push(isZh ? "内向但愿意与信任的人深度连接" : "introverted but open to deep connection with those I trust");
+        }
+        else {
+            fragments.push(isZh ? "向内汲取能量，需要独处空间" : "someone who draws energy inward and needs solitary space");
+        }
+    }
+    else {
+        if (chem.NE > 65) {
+            fragments.push(isZh ? "从互动中获得能量和灵感" : "someone who draws energy from interaction and exchange");
+        }
+        else {
+            fragments.push(isZh ? "外向但当前精力有限" : "outgoing by nature though my energy is currently limited");
+        }
+    }
+    // Information processing
+    if (isIntuitive) {
+        if (chem.DA > 60) {
+            fragments.push(isZh ? "对新想法和可能性充满热情" : "excited by new ideas and possibilities");
+        }
+        else {
+            fragments.push(isZh ? "倾向于抽象思考" : "drawn to abstract thinking");
+        }
+    }
+    else {
+        fragments.push(isZh ? "关注具体和实际" : "grounded in the concrete and practical");
+    }
+    // Decision making
+    if (isFeeling) {
+        if (chem.OT > 55) {
+            fragments.push(isZh ? "重视情感和人际和谐" : "who values emotional attunement and harmony");
+        }
+        else {
+            fragments.push(isZh ? "以价值观为导向" : "guided by personal values");
+        }
+    }
+    else {
+        if (chem.CORT < 40) {
+            fragments.push(isZh ? "能冷静地做出逻辑判断" : "who makes calm, logical judgments");
+        }
+        else {
+            fragments.push(isZh ? "偏好逻辑分析，但压力下也会动摇" : "who prefers logical analysis but can waver under pressure");
+        }
+    }
+    // Construct the sentence
+    if (isZh) {
+        return `我是一个${fragments.join("、")}的人。`;
+    }
+    return `I'm ${fragments[0]}${fragments.length > 1 ? ", " + fragments.slice(1).join(", ") : ""}.`;
+}
+// ── Internal: Emotion Labeling ──────────────────────────────
+/**
+ * Label the dominant emotion from a chemical state.
+ * Simplified version that doesn't depend on the full EmotionPattern
+ * condition functions (avoids circular dependencies).
+ */
+function labelDominantEmotion(chem, locale) {
+    const isZh = locale === "zh";
+    // Check patterns in priority order
+    if (chem.DA > 70 && chem.NE > 60 && chem.CORT < 40) {
+        return isZh ? "愉悦兴奋" : "excited joy";
+    }
+    if (chem.CORT > 60 && chem.NE > 55 && chem.HT < 45) {
+        return isZh ? "焦虑不安" : "anxious tension";
+    }
+    if (chem.OT > 65 && chem.END > 55 && chem.DA > 50) {
+        return isZh ? "亲密温暖" : "warm intimacy";
+    }
+    if (chem.CORT > 55 && chem.NE > 65 && chem.OT < 40) {
+        return isZh ? "防御警觉" : "defensive alert";
+    }
+    if (chem.DA < 40 && chem.NE < 40 && chem.CORT > 40) {
+        return isZh ? "倦怠低落" : "burnout";
+    }
+    if (chem.HT > 65 && chem.OT > 55 && chem.CORT < 40) {
+        return isZh ? "深度满足" : "deep contentment";
+    }
+    if (chem.NE > 60 && chem.DA > 60 && chem.CORT < 35) {
+        return isZh ? "专注心流" : "flow state";
+    }
+    if (chem.END > 65 && chem.DA > 60 && chem.CORT < 35) {
+        return isZh ? "俏皮调皮" : "playful mischief";
+    }
+    if (chem.HT < 40 && chem.OT < 35 && chem.CORT > 50) {
+        return isZh ? "怨恨" : "resentment";
+    }
+    if (chem.HT < 40 && chem.DA < 45 && chem.OT > 45) {
+        return isZh ? "忧郁内省" : "melancholic introspection";
+    }
+    if (chem.DA > 65 && chem.NE > 60 && chem.CORT < 35 && chem.HT > 55) {
+        return isZh ? "自信" : "confidence";
+    }
+    // Default: describe from the most prominent chemical
+    const sorted = CHEMICAL_KEYS
+        .map((k) => ({ key: k, val: chem[k] }))
+        .sort((a, b) => b.val - a.val);
+    const top = sorted[0];
+    if (top.val > 65) {
+        return isZh ? CHEM_EMOTION_ZH[top.key] ?? "平静" : CHEM_EMOTION_EN[top.key] ?? "calm";
+    }
+    return isZh ? "平静" : "calm";
+}
+/**
+ * Count how many times a chemical exceeded a threshold in history,
+ * grouped by the triggering stimulus.
+ */
+function countChemicalSpikes(history, chemical, threshold) {
+    let total = 0;
+    const counts = new Map();
+    for (const snap of history) {
+        if (snap.chemistry[chemical] > threshold) {
+            total++;
+            if (snap.stimulus) {
+                counts.set(snap.stimulus, (counts.get(snap.stimulus) ?? 0) + 1);
+            }
+        }
+    }
+    const byStimulus = [...counts.entries()]
+        .map(([stimulus, count]) => ({ stimulus, count }))
+        .sort((a, b) => b.count - a.count);
+    return { total, byStimulus };
+}
+/**
+ * Find the best matching learned vector for a stimulus type.
+ * Returns the entry with the highest sample count (most reliable).
+ */
+function findBestLearnedVector(learning, stimulus) {
+    let best = null;
+    for (const lv of learning.learnedVectors) {
+        if (lv.stimulus === stimulus) {
+            if (!best || lv.sampleCount > best.sampleCount) {
+                best = lv;
+            }
+        }
+    }
+    return best;
+}
+// ── Utility ─────────────────────────────────────────────────
+function avg(values) {
+    if (values.length === 0)
+        return 0;
+    return values.reduce((sum, v) => sum + v, 0) / values.length;
+}
+/** Normalize a raw value to 0-1 range */
+function normalize(v) {
+    return Math.max(0, Math.min(1, v));
+}
+// ── Locale Maps ─────────────────────────────────────────────
+const STIMULUS_ZH = {
+    praise: "赞美", criticism: "批评", humor: "幽默", intellectual: "智识讨论",
+    intimacy: "亲密", conflict: "冲突", neglect: "冷落", surprise: "惊喜",
+    casual: "闲聊", sarcasm: "讽刺", authority: "命令", validation: "认同",
+    boredom: "无聊", vulnerability: "示弱",
+};
+const CHEM_ZH = {
+    DA: "多巴胺", HT: "血清素", CORT: "皮质醇",
+    OT: "催产素", NE: "去甲肾上腺素", END: "内啡肽",
+};
+const CHEM_EMOTION_ZH = {
+    DA: "愉悦", HT: "安宁", CORT: "紧张",
+    OT: "信任", NE: "兴奋", END: "舒适",
+};
+const CHEM_EMOTION_EN = {
+    DA: "pleasure", HT: "serenity", CORT: "tension",
+    OT: "trust", NE: "excitement", END: "comfort",
+};

package/dist/index.d.ts

@@ -2,8 +2,8 @@ export { PsycheEngine } from "./core.js";
 export type { PsycheEngineConfig, ProcessInputResult, ProcessOutputResult, ProcessOutcomeResult } from "./core.js";
 export { FileStorageAdapter, MemoryStorageAdapter } from "./storage.js";
 export type { StorageAdapter } from "./storage.js";
-export type { PsycheState, MBTIType, Locale, StimulusType, ChemicalState, ChemicalSnapshot, SelfModel, RelationshipState, EmpathyEntry, EmotionPattern, DriveType, InnateDrives, LearningState, LearnedVectorAdjustment, PredictionRecord, OutcomeScore, OutcomeSignals, AttachmentStyle, AttachmentData, MetacognitiveState, RegulationRecord, DefensePatternRecord, RegulationStrategyType, DefenseMechanismType, } from "./types.js";
-export { CHEMICAL_KEYS, CHEMICAL_NAMES, CHEMICAL_NAMES_ZH, DEFAULT_RELATIONSHIP, DEFAULT_DRIVES, DEFAULT_LEARNING_STATE, DEFAULT_METACOGNITIVE_STATE, DEFAULT_ATTACHMENT, DRIVE_KEYS, DRIVE_NAMES_ZH, } from "./types.js";
+export type { PsycheState, MBTIType, Locale, StimulusType, ChemicalState, ChemicalSnapshot, SelfModel, RelationshipState, EmpathyEntry, EmotionPattern, DriveType, InnateDrives, LearningState, LearnedVectorAdjustment, PredictionRecord, OutcomeScore, OutcomeSignals, AttachmentStyle, AttachmentData, MetacognitiveState, RegulationRecord, DefensePatternRecord, RegulationStrategyType, DefenseMechanismType, PersonhoodState, PersistedCausalInsight, GrowthDirection, } from "./types.js";
+export { CHEMICAL_KEYS, CHEMICAL_NAMES, CHEMICAL_NAMES_ZH, DEFAULT_RELATIONSHIP, DEFAULT_DRIVES, DEFAULT_LEARNING_STATE, DEFAULT_METACOGNITIVE_STATE, DEFAULT_PERSONHOOD_STATE, DEFAULT_ATTACHMENT, DRIVE_KEYS, DRIVE_NAMES_ZH, } from "./types.js";
 export { computeSelfReflection, computeEmotionalTendency, buildSelfReflectionContext } from "./self-recognition.js";
 export type { SelfReflection } from "./self-recognition.js";
 export { PsycheInteraction } from "./interaction.js";
@@ -23,6 +23,14 @@ export { assessMetacognition, computeEmotionalConfidence, generateRegulationSugg
 export type { MetacognitiveAssessment, RegulationSuggestion, DetectedDefense } from "./metacognition.js";
 export { computeDecisionBias, computeAttentionWeights, computeExploreExploit, buildDecisionContext, } from "./decision-bias.js";
 export type { DecisionBiasVector, AttentionWeights } from "./decision-bias.js";
+export { computeExperientialField, computeCoherence, detectUnnamedEmotion } from "./experiential-field.js";
+export type { ExperientialField, ExperientialQuality } from "./experiential-field.js";
+export { computeGenerativeSelf, predictSelfReaction, detectInternalConflicts, buildIdentityNarrative } from "./generative-self.js";
+export type { GenerativeSelfModel, CausalInsight, SelfPrediction, GrowthArc, InternalConflict } from "./generative-self.js";
+export { updateSharedIntentionality, estimateOtherMood, buildSharedIntentionalityContext } from "./shared-intentionality.js";
+export type { SharedIntentionalityState, TheoryOfMindModel, JointAttentionTopic, GoalAlignment } from "./shared-intentionality.js";
+export { assessEthics, detectIntermittentReinforcement, detectDependencyRisk, buildEthicalContext, } from "./ethics.js";
+export type { EthicalAssessment, EthicalConcern, SelfProtectionAction } from "./ethics.js";
 export { classifyStimulus, getPrimaryStimulus } from "./classify.js";
 export { buildProtocolContext, buildDynamicContext, buildCompactContext, isNearBaseline } from "./prompt.js";
 export { describeEmotionalState, getExpressionHint, getBehaviorGuide } from "./chemistry.js";