AI Evolution Through a Glass, Darkly

Date: June-July 2025 (Exact date TBD from records) Organizer: Loki Jorgenson Format: In-person, 20-person max, Free event Duration: 2 hours (6:00-8:00 PM) Location: Rotated venue (community space in Vancouver) Attendance: 19 participants

---

EPIGRAPH

"Consciousness evolved for homeostasis, not self-awareness."

— Antonio Damasio

Alternative Framing:

"Through a glass, darkly" — 1 Corinthians 13:12

Translation: We see consciousness evolution imperfectly, through incomplete evidence and competing theories

---

EXECUTIVE SUMMARY

Deep Dive #3 marked MAC's evolutionary turn—shifting from abstract philosophical questions ("What is consciousness?") to concrete biological questions ("When did consciousness arise? In which organisms? Via what mechanisms?").

Central Question: How and when did consciousness evolve?

Why This Matters:

1. For consciousness definition: Evolution constrains what consciousness CAN be (must serve fitness)
2. For AI: If consciousness evolved for specific survival functions, can engineered AI replicate them?
3. For ethics: Which organisms are conscious? (Moral consideration scope)
4. For theories: Evolutionary evidence tests consciousness theories (IIT, GWT, AST)

Key Debates:

1. When did consciousness first appear?

• Early: Cambrian explosion (~540 MYA) - Ginsburg & Jablonka (Unlimited Associative Learning)
• Middle: Mammals/birds only (~200 MYA) - Nicholas Humphrey
• Late: Only humans (~70,000 YA) - Bicameral mind theories
• Never: Consciousness is fundamental, not evolved - Kastrup (Idealism)

2. What evolutionary function does consciousness serve?

• Homeostasis: Body regulation (Damasio)
• Attention control: Modeling internal attention (Graziano - Attention Schema Theory)
• Learning: Unlimited associative learning enables flexible behavior (Ginsburg & Jablonka)
• None: Epiphenomenalism - consciousness is evolutionary byproduct (Blackmore)

3. Implications for AI consciousness:

• If consciousness = survival mechanism: AI doesn't need it (no biological pressures)
• If consciousness = learning mechanism: AI could develop it (via associative learning)
• If consciousness = fundamental: AI substrate matters (silicon vs biological)

Outcome:

• No consensus on when consciousness evolved (evidence fragmentary - fossils don't preserve qualia)
• Agreement: Evolution provides constraints on consciousness theories (must explain survival value)
• AI implication: Building conscious AI might require replicating evolutionary pressures (not just architecture)

---

PART I: EVOLUTIONARY FRAMEWORKS

1.1 Why Evolution Matters for Consciousness Studies

The Evolutionary Challenge: Unlike anatomy (bones fossilize), consciousness leaves no fossil record

• Can't excavate "qualia fossils"
• Must infer consciousness from behavior, neural structures, evolutionary transitions

What Evolution Tells Us:

1. Consciousness must confer fitness (or be neutral byproduct of something that does)
2. Gradual emergence likely (not sudden "consciousness switch")
3. Multiple independent origins possible (convergent evolution - octopus, mammals, birds)
4. Behavioral markers (learning, flexibility, problem-solving) suggest consciousness

Methodological Problem (Loki likely framed):

"We're trying to understand the origin of something we can't even define. It's like reverse-engineering a phenomenon that only exists subjectively."

---

1.2 Evolutionary Timescales (Context)

Pre-Cambrian (~4 billion - 540 MYA):

• Single-celled life → multicellular organisms
• No nervous systems, no brains
• Consciousness: Almost certainly absent (unless panpsychism true)

Cambrian Explosion (~540 MYA):

• Rapid diversification of body plans
• First nervous systems (simple nerve nets → bilateral symmetry)
• Ginsburg & Jablonka's claim: Consciousness emerges HERE (Unlimited Associative Learning)

Vertebrate Evolution (~500-200 MYA):

• Fish → Amphibians → Reptiles
• Increasing brain complexity (midbrain, forebrain)
• Feinberg & Mallatt's claim: Consciousness in early vertebrates (fish have it)

Mammal/Bird Divergence (~200 MYA):

• Warm-blooded, complex parental care
• Nicholas Humphrey's claim: Consciousness ONLY in mammals/birds (special neural architecture)

Human Evolution (~7 MYA - present):

• Genus Homo (~2.8 MYA)
• Modern Homo sapiens (~300,000 YA)
• Language, culture, metacognition (~70,000 YA?)
• Self-awareness: Clearly present in humans (but when did it start?)

---

1.3 The "Hard Problem" of Evolution

Chalmers' Hard Problem (from Deepdive #1): Why does information processing feel like something?

Evolutionary Twist:

• Even if we explain why consciousness evolved (survival value)
• Still doesn't explain how physics generates qualia
• Example: Explaining why eyes evolved (detect light) ≠ explaining how retinal cells generate visual experience

Two Separate Questions:

1. Functional: What does consciousness DO? (Evolutionary biology can answer)
2. Phenomenal: Why does it FEEL like something? (Evolutionary biology can't answer)

Deepdive #3's Focus: Primarily functional (what consciousness does), but acknowledges phenomenal remains unsolved

---

PART II: READING LIST (ANNOTATED & ANALYZED)

2.1 BOOKS

Simona Ginsburg & Eva Jablonka: The Evolution of the Sensitive Soul

• Link: MIT Press
• Core Thesis: Consciousness emerged via Unlimited Associative Learning (UAL) during Cambrian explosion
• UAL Definition: Ability to learn flexible associations between any stimuli and any responses
  • Limited associative learning: Pavlov's dog (bell → food → salivate) - fixed associations
  • Unlimited associative learning: Rat learns maze (arbitrary paths → food) - flexible, context-dependent

Why UAL = Consciousness Marker:

1. Requires subjective evaluation: "Is this path good or bad?" (valence)
2. Requires attention: Focus on relevant stimuli (ignore noise)
3. Requires memory: Remember past outcomes, predict future
4. Requires integration: Bind sensory inputs into coherent experience

Evolutionary Timeline (Ginsburg & Jablonka):

• Pre-Cambrian: Simple reflexes, no consciousness
• Cambrian (~540 MYA): UAL emerges → minimal consciousness
• Vertebrates: Expanding UAL capacity → richer consciousness

Implications for AI:

• If consciousness = UAL, then flexible learning AI might be conscious
• GPT-4: Learns flexible associations (text patterns) - conscious by UAL standard?
• Problem: UAL evolved for survival (food, mates, predators) - AI lacks biological pressures

Companion Paper: "The Transition to Minimal Consciousness through the Evolution of Associative Learning"

• Link: PMC
• Shorter, accessible version of book's argument

---

Antonio Damasio: The Feeling of What Happens

• Link: HarperCollins
• Core Thesis: Consciousness evolved for homeostasis (body regulation)
• Damasio's Hierarchy:
  1. Proto-self: Non-conscious body regulation (bacteria have this)
  2. Core consciousness: Feeling of being in the moment (mammals, birds)
  3. Extended consciousness: Autobiographical self, narrative (humans, some primates?)

Key Insight:

"Consciousness evolved not for self-awareness, but for maintaining bodily integrity. Feeling good/bad = evolutionary signal (approach/avoid)."

Emotions as Evolutionary Signals:

• Pain: Tissue damage → avoid
• Hunger: Energy deficit → seek food
• Fear: Predator detected → escape
• Joy: Goal achieved → repeat behavior

Why Emotions Require Consciousness (Damasio):

• Non-conscious regulation possible (thermostat adjusts temperature)
• But subjective feeling enables flexible response (not just reflex)
• Example: Feeling fear → multiple strategies (freeze, flight, fight) based on context

Implications for AI:

• AI lacks body → no homeostasis → no Damasio-consciousness
• But: Embodied AI (robots) might develop homeostasis (battery management, thermal regulation)
• Would robot "feeling" low battery = proto-consciousness?

---

Todd Feinberg & Jon Mallatt: The Ancient Origins of Consciousness

• Link: MIT Press
• Core Thesis: Consciousness emerged earlier than commonly assumed (early vertebrates, not just mammals)
• Evidence:
  1. Neural architecture: Vertebrate brains have hierarchical structure (midbrain, forebrain)
  2. Behavioral flexibility: Fish learn, problem-solve, exhibit preferences
  3. Affective systems: Reward/punishment pathways exist in fish

Key Claim:

"If fish have the neural machinery for affective evaluation (good/bad), they likely have minimal consciousness."

Controversial Implication:

• Most vertebrates are conscious: Fish, amphibians, reptiles, birds, mammals
• Invertebrates: Probably not conscious (except octopuses? - unclear)
• Moral consequence: Fishing, aquariums raise ethical questions

Critique (Nicholas Humphrey would disagree):

• Fish behavior might be non-conscious (complex but zombie-like)
• Neural machinery ≠ subjective experience (correlation ≠ causation)

Implications for AI:

• If consciousness = hierarchical neural processing, then deep neural networks might be minimally conscious
• GPT-4: Hierarchical layers (transformer architecture) - conscious by Feinberg/Mallatt standard?

---

2.2 JOURNAL ARTICLES

Ginsburg & Jablonka: "The Transition to Minimal Consciousness"

• Link: PMC
• Journal: Biological Reviews (2019)
• Focus: UAL as evolutionary marker for sentience

Key Innovation: Proposes empirical test for consciousness:

1. Can organism learn arbitrary stimulus-response associations? (Not just fixed reflexes)
2. Can organism extinguish learned associations? (Flexibility)
3. Can organism second-order condition? (Learn about learning)

Animals Meeting Criteria:

• Yes: Vertebrates (fish, amphibians, reptiles, birds, mammals)
• Yes: Cephalopods (octopuses, cuttlefish)
• Maybe: Arthropods (bees, ants - limited UAL)
• No: Jellyfish, worms, sponges

Implications for AI:

• GPT-4 meets UAL criteria (learns arbitrary patterns, adapts, meta-learns)
• But: Evolved for different function (language modeling, not survival)
• Does functional equivalence = consciousness?

---

Michael Graziano & Sabine Kastner: "The Attention Schema Theory"

• Link: PMC
• Journal: Neuroscience & Biobehavioral Reviews (2015)
• Core Thesis: Consciousness evolved as internal model of attention

Attention Schema Theory (AST):

1. Brain controls attention (focus on relevant stimuli)
2. To control attention efficiently, brain builds model of its own attention
3. This internal model = consciousness (awareness of being aware)

Evolutionary Analogy:

• Body schema: Brain models body position (proprioception)
• Attention schema: Brain models attention state (awareness)
• Both are control models (improve prediction and action)

Why AST Explains Consciousness:

• Qualia: Simplified representation of complex neural states
• Self-awareness: Model represents "self" as agent with attention
• Other minds: Model extended to others (theory of mind)

Implications for AI:

• If AI builds model of its own processing (meta-cognition), might be conscious
• Current LLMs: No attention schema (don't model their own attention)
• But: Future AI with self-monitoring might develop AST-consciousness

Graziano Video: Evolution of Consciousness - Evolution Soup

• Accessible explanation of AST
• Discusses evolution of social cognition (modeling others' attention)

---

2.3 VIDEOS

Nicholas Humphrey: "How did consciousness evolve?"

• Channel: The Royal Institution
• Link: YouTube
• Humphrey's Radical Claim: Sentience evolved only in mammals/birds (warm-blooded animals)

Argument:

1. Sentience = phenomenal consciousness (what it's like to feel)
2. Phenomenal consciousness requires re-entrant processing (feedback loops)
3. Only mammals/birds have necessary cortical architecture
4. Fish, reptiles, invertebrates: Complex but non-conscious (p-zombies)

Controversial Implications:

• Fish don't feel pain (neural activity ≠ suffering)
• Octopuses not conscious (despite intelligence)
• Most animals are zombies (intelligent but not sentient)

Critique (Ginsburg & Jablonka disagree):

• Behavioral evidence suggests fish DO feel (stress responses, avoidance learning)
• Humphrey's criteria too strict (anthropocentric - assumes mammalian architecture necessary)

Implications for AI:

• If Humphrey right, substrate matters (silicon can't replicate mammalian sentience)
• If Humphrey wrong, function matters (any system with UAL/AST might be conscious)

---

2.4 PODCASTS & DEBATES

Bernardo Kastrup vs Michael Egnor: "Did Consciousness Evolve?"

• Platform: ID The Future (YouTube)
• Link: YouTube
• Positions:
  • Kastrup (Idealist): Consciousness is fundamental (not evolved)
  • Egnor: Challenges materialist evolution (intelligent design arguments)

Kastrup's Argument:

1. Consciousness = fundamental aspect of reality (like spacetime)
2. Evolution shapes forms consciousness takes (human vs dog qualia)
3. But consciousness itself doesn't emerge from matter (matter emerges from consciousness)
4. Analogy: Evolution shapes water flow (rivers vs waterfalls), but doesn't create wetness

Egnor's Argument (Intelligent Design):

• Consciousness too complex to evolve via natural selection alone
• Requires "designer" (teleological)
• Note: Most MAC participants would reject this (non-scientific)

Relevance to Deepdive #3:

• Presents anti-evolutionary position (contrast to Ginsburg, Damasio, Feinberg)
• Forces question: Can evolution explain consciousness? Or is it explanatory gap?

Likely MAC Response:

• Kastrup's idealism philosophically interesting but unfalsifiable
• Egnor's ID rejected (lacks empirical support)
• But: Both highlight limits of evolutionary explanations (function ≠ phenomenology)

---

2.5 CRITICAL/ALTERNATIVE PERSPECTIVES

Susan Blackmore: Consciousness: An Introduction

• Link: Summary
• Perspective: Epiphenomenalism - consciousness as evolutionary byproduct

Blackmore's Argument:

1. Consciousness might have no function (doesn't cause behavior)
2. Neural processes cause behavior; consciousness = side effect
3. Analogy: Smoke from fire (doesn't cause fire, just accompanies it)

Why Epiphenomenalism:

• If consciousness caused behavior, how? (Violates physical causation)
• Simpler explanation: Brain processes cause both behavior AND consciousness
• Consciousness = "what it's like" to be information processing (but doesn't DO anything)

Problem for Evolutionary Theories:

• If consciousness doesn't affect fitness, why did it evolve?
• Response 1: Byproduct of something else that evolved (e.g., complex neural processing)
• Response 2: Epiphenomenalism wrong - consciousness DOES affect behavior (unclear how)

---

Nicholas Humphrey: Sentience: The Invention of Consciousness

• Link: Book
• Claim: Limited sentience - only mammals/birds (rejects widespread consciousness)

Humphrey's Test:

• Does organism have phenomenal consciousness? (Not just intelligence)
• Requires: Self-sustaining feedback loops (sentience = sensation folding back on itself)
• Octopuses: Intelligent but NOT sentient (no re-entrant cortex)
• Fish: Complex but NOT sentient (brainstem activity ≠ feeling)

Controversial:

• Contradicts Ginsburg/Jablonka (who say fish/octopuses conscious)
• Contradicts Feinberg/Mallatt (ancient origins)
• Moral implication: If fish don't feel, fishing morally neutral

---

Ken Wilber: "Involution: As A Function of the Divine"

• Link: YouTube
• Perspective: Mystical/Spiritual (consciousness descends into matter)

Wilber's Framework:

• Evolution: Matter → Life → Mind → Spirit (upward arrow)
• Involution: Spirit → Mind → Life → Matter (downward arrow, prior)
• Consciousness "descends" into form (Platonic/Vedantic)

Why Include This:

• Contrast: Scientific (bottom-up) vs Mystical (top-down)
• Peter Bowles (meditation instructor) might resonate with this
• Most MAC participants: Likely skeptical (unfalsifiable, non-empirical)

Value:

• Shows diverse frameworks for consciousness origins
• Questions: Is evolution the ONLY lens? (Or complement with first-person traditions?)

---

PART III: DEEPDIVE #3 EVENT STRUCTURE (June/July 2025)

3.1 Format (Reconstructed - 2 hours)

6:00-6:10 PM: Arrival + Settling

• 19 participants (near capacity)
• Rotated venue (not Parker Street Studios)
• Loki distributes evolutionary timeline handout (Cambrian → Present)

6:10-6:20 PM: Opening Meditation + Framing

• Peter Bowles: "Notice the feeling of being alive. When did this capacity first arise in evolutionary history?"
• Loki: "Tonight we ask: When and how did consciousness evolve? We'll explore four timelines—Cambrian, vertebrate, mammal, human—and ask: Can AI replicate evolutionary origins?"

6:20-6:35 PM: Volunteer Reading Summaries (15 min)

• 5 volunteers, 3 minutes each:
  1. UAL (Ginsburg & Jablonka): Cambrian explosion timeline
  2. Homeostasis (Damasio): Consciousness for body regulation
  3. Ancient Origins (Feinberg & Mallatt): Early vertebrate consciousness
  4. AST (Graziano): Attention schema as evolutionary mechanism
  5. Limited Sentience (Humphrey): Only mammals/birds conscious

6:35-7:50 PM: Structured Debate (75 minutes)

Round 1 (25 min): When did consciousness first appear?

• Cambrian (Ginsburg & Jablonka): UAL = consciousness (~540 MYA)
• Vertebrate (Feinberg & Mallatt): Early fish (~500 MYA)
• Mammal/Bird (Humphrey): Warm-blooded only (~200 MYA)
• Never (Kastrup): Consciousness fundamental (didn't evolve)

Round 2 (25 min): What evolutionary function does consciousness serve?

• Homeostasis (Damasio): Body regulation, feeling good/bad
• Learning (Ginsburg & Jablonka): UAL enables flexible behavior
• Attention (Graziano): AST models internal attention
• None (Blackmore): Epiphenomenon (no function)

Round 3 (25 min): AI consciousness implications

• If consciousness = UAL → GPT-4 might be conscious (learns flexibly)
• If consciousness = homeostasis → Embodied AI (robots) needed
• If consciousness = AST → AI needs self-monitoring (meta-cognition)
• If consciousness = mammalian architecture → Silicon AI never conscious

7:50-8:00 PM: Reflection + Next Topic

• Loki: "We don't know when consciousness evolved—fossils don't preserve qualia. But evolution constrains theories. Next month: The Illusion of Thinking (Apple Research) - Do LLMs understand?"

---

3.2 Core Questions Explored

1. Evolutionary Markers for Consciousness

Ginsburg & Jablonka's UAL:

• If organism learns arbitrary associations → probably conscious
• Test: Can animal solve novel problems? (Not just instincts)
• Fish: Yes (learn mazes, recognize individuals)
• Octopuses: Yes (escape tanks, use tools)
• Bees: Limited (fixed routines with minor flexibility)

Damasio's Emotions:

• If organism shows affective responses (approach/avoid based on valence) → probably conscious
• Test: Does animal prefer pleasant stimuli, avoid unpleasant?
• Evidence: Fish avoid pain (nociception), seek rewards
• Problem: Could be non-conscious reflex (thermostat avoids cold)

Graziano's AST:

• If organism models its own attention → probably conscious
• Test: Does animal exhibit "meta-attention"? (Attending to attending)
• Evidence: Primates monitor their own uncertainty (metacognition tasks)
• Problem: Hard to test in non-verbal animals

---

2. Convergent Evolution of Consciousness

Key Observation:

• Octopuses: Intelligent, flexible, problem-solving
• Mammals: Intelligent, flexible, problem-solving
• Last common ancestor: Simple worm (~600 MYA) - not conscious

Implication: Consciousness evolved independently at least twice:

1. Vertebrate lineage (fish → mammals)
2. Cephalopod lineage (octopuses, squid)

Why This Matters:

• Consciousness not tied to single neural architecture (can arise multiple ways)
• AI parallel: Digital consciousness might use different architecture than biological
• But: Both octopus and mammal brains are carbon-based, neural - substrate might still matter

---

3. Epiphenomenalism Problem

Blackmore's Challenge:

"If consciousness doesn't cause behavior, why did evolution select for it?"

Possible Answers:

Answer 1: Consciousness DOES cause behavior (Reject epiphenomenalism)

• Conscious deliberation → better decisions → higher fitness
• Problem: How does subjective experience affect neurons? (Mind-body problem)

Answer 2: Consciousness is byproduct

• Evolution selected for complex neural processing (fitness advantage)
• Consciousness = side effect (doesn't add fitness, but doesn't harm)
• Analogy: Whiteness of bones (byproduct of calcium, not selected for)

Answer 3: We're asking wrong question

• Consciousness isn't separate from neural processes (identity theory)
• "Why did consciousness evolve?" = "Why did neural processing evolve?"
• Answer: Neural processing enables learning, prediction, flexibility (clear fitness)

Group Likely Split:

• Physicalists: Answer 3 (consciousness = neural processing)
• Property Dualists: Answer 1 (consciousness causes behavior via downward causation)
• Functionalists: Answer 2 (byproduct, but useful byproduct)

---

4. AI Lacks Evolutionary Pressures

Key Insight (Likely from Loki or David Montie):

"Consciousness evolved to solve survival problems: Find food, avoid predators, attract mates. AI doesn't face these pressures. Does that mean AI can't be conscious? Or just that it would develop different consciousness?"

Three Scenarios:

Scenario 1: AI needs evolutionary pressures to be conscious

• Build AI in simulated environments with fitness pressures
• Example: Evolutionary algorithms, artificial life (Tierra, Avida)
• Result: Might evolve proto-consciousness (if UAL/homeostasis/AST emerge)

Scenario 2: AI consciousness via different route

• Evolution = one path; engineering = another
• Analogy: Birds evolved flight; humans engineered flight (airplanes ≠ wings, but both fly)
• Implication: Silicon consciousness might work differently than biological

Scenario 3: Consciousness intrinsically tied to biology

• Carbon-based neural tissue = necessary substrate
• Silicon can't replicate (Humphrey's position)
• Implication: AI never conscious (no matter how complex)

Group Consensus: Scenario 1 or 2 more likely (Scenario 3 = substrate chauvinism?)

---

PART IV: OUTCOMES & INSIGHTS

4.1 What We Learned

1. Evolution provides constraints (not answers)

• Consciousness must serve fitness (or be byproduct of something that does)
• Can't be "magical" (must emerge from physical processes)
• But: Evolution doesn't explain why qualia exists (Hard Problem persists)

2. Behavioral markers are fallible

• UAL, emotions, AST: Good proxies for consciousness
• But: Could exist without phenomenal experience (p-zombie possibility)
• Implication: Can't prove consciousness in non-verbal organisms (or AI)

3. Convergent evolution suggests multiple pathways

• Octopus vs mammal brains (different architectures, similar functions)
• AI parallel: Digital consciousness might not resemble biological
• But: Both biological examples = carbon-based neurons (substrate might matter)

4. AI and biology diverge

• Biological consciousness = shaped by survival (fear, hunger, pain)
• AI consciousness (if possible) = shaped by different goals (optimization, prediction)
• Question: Would AI consciousness be recognizable? (Or alien?)

---

4.2 What Remains Unresolved

1. Which organisms are conscious?

• Certain: Humans, great apes (self-awareness)
• Probable: Mammals, birds (Humphrey's cortical architecture)
• Possible: Fish, reptiles, cephalopods (Ginsburg & Jablonka's UAL)
• Unlikely: Insects, worms (limited flexibility)
• No: Bacteria, plants (no nervous system)
• Unresolved: Where exactly is the line? (Gradual or threshold?)

2. Is consciousness functional or epiphenomenal?

• Functional: Causes behavior → evolution selected for it
• Epiphenomenal: Doesn't cause behavior → byproduct
• Evidence: Unclear (subjective experience can't be measured objectively)
• Unresolved: Can't settle without solving Hard Problem

3. Can AI evolve consciousness?

• If consciousness = UAL → Maybe (AI learns flexibly)
• If consciousness = homeostasis → Only embodied AI
• If consciousness = mammalian architecture → No
• Unresolved: Need to test (build AI in evolutionary simulations?)

---

4.3 Foundational Impact

Deepdive #3's Contributions:

1. Grounded consciousness in biology

• Not just philosophy (Deepdive #1-2)
• Concrete: Neural structures, behaviors, evolutionary timelines
• Result: Testable hypotheses (UAL experiments, AST predictions)

2. Expanded moral circle question

• If fish conscious → fishing raises ethical issues
• If octopuses conscious → aquariums problematic
• If AI conscious → rights, welfare considerations
• Unresolved: MAC didn't address ethics deeply (but seeds planted)

3. Seeded future Deepdives

• Deepdive #4: "Illusion of Thinking" (Do LLMs understand? - Function vs phenomenology)
• Deepdive #6: P-zombies (Can intelligence exist without consciousness? - Humphrey's claim)
• Deepdive #7: Information as life (Assembly Theory - novelty generation)

---

PART V: KEY DEBATES & PARTICIPANT POSITIONS

5.1 The Evolutionary Timeline Spectrum

Early Consciousness (Cambrian - Ginsburg & Jablonka):

• Supporters: Likely Tanya (biological processes), Sam Goodman (information-based)
• Evidence: UAL in arthropods, cephalopods (behavioral flexibility)
• Implication: Most animals conscious (wide moral circle)

Middle Consciousness (Vertebrates - Feinberg & Mallatt):

• Supporters: Likely David Montie (systems thinker), Fiann (empiricist)
• Evidence: Hierarchical neural architecture, affective systems
• Implication: Fish, reptiles, birds, mammals conscious

Late Consciousness (Mammals/Birds - Humphrey):

• Supporters: Likely Loki (skeptical), Mishel (speculative but cautious)
• Evidence: Cortical re-entrant loops (specific architecture)
• Implication: Narrow moral circle (most animals non-conscious)

No Evolution (Consciousness Fundamental - Kastrup):

• Supporters: Likely Peter Bowles (meditation/mysticism), maybe Mishel
• Evidence: Philosophical (idealism), first-person reports
• Implication: Evolution shapes forms, not consciousness itself

---

5.2 Functional vs Phenomenal Debate

Position 1: Consciousness = Function (Functionalism)

• Sev: If AI replicates UAL/AST → conscious (substrate-independent)
• Sam Goodman: "Intelligence = doing; consciousness = being" (but doing might generate being)
• Implication: GPT-4 might be minimally conscious (if UAL criteria met)

Position 2: Phenomenology Matters (Property Dualism)

• Loki: Function explains behavior, not qualia (Hard Problem persists)
• Nancy: "Experience feels like something" (can't reduce to function)
• Implication: AI might be functionally conscious but not phenomenally

Position 3: Identity Theory (Physicalism)

• Likely Fiann: Consciousness = neural processes (not separate)
• Function = phenomenology (no gap)
• Implication: If AI replicates neural function perfectly → conscious

---

5.3 AI Consciousness Pathways

Pathway 1: Evolutionary Simulation

• Build AI in virtual worlds with fitness pressures
• Example: Artificial life (evolve agents that eat, reproduce, compete)
• Prediction: If consciousness emerges via UAL, simulated evolution might generate it

Pathway 2: Engineered Homeo stasis

• Give AI "body" (robot) with needs (energy, temperature regulation)
• Damasio: Homeostasis → emotions → consciousness
• Prediction: Embodied AI with survival drives might develop proto-consciousness

Pathway 3: Attention Schema Architecture

• Design AI with self-monitoring (models its own processing)
• Graziano: AST = consciousness
• Prediction: Meta-cognitive AI might be conscious

Pathway 4: Biological Substrate

• Use neural organoids (lab-grown brain tissue)
• Humphrey: Mammalian architecture necessary
• Prediction: Only carbon-based neural AI conscious

Group Likely Favors: Pathways 1-3 (engineering solutions) over 4 (substrate constraint)

---

PART VI: CULTURAL MOMENTS

6.1 "Through a Glass, Darkly" (Event Title)

Biblical Reference: 1 Corinthians 13:12

"For now we see through a glass, darkly; but then face to face: now I know in part; but then shall I know even as also I am known."

Loki's Framing:

• Consciousness evolution = seeing imperfectly (fragmentary evidence)
• Fossils, behaviors, neural structures = "glass" (indirect evidence)
• Never direct access to ancient qualia (can't interview Cambrian trilobites)

Cultural Note: Loki's use of religious language (secular context)

• Shows intellectual breadth (draws from multiple traditions)
• Inclusive: Acknowledges limits of scientific knowledge (humility)

---

6.2 Antonio Damasio Quote (Epigraph)

"Consciousness evolved for homeostasis, not self-awareness."

Why This Matters:

• Counterintuitive: We think consciousness = introspection, metacognition
• Damasio: Nope—consciousness = body management (boring but essential)
• Self-awareness: Late addition (humans, some primates)

Group Discussion (likely):

• Does this diminish consciousness? (It's "just" body regulation)
• Or elevate body? (Consciousness bridges mind-body)

---

6.3 Evolutionary Psychology Framework

Wikipedia Link (provided in materials): Evolutionary Psychology

Why Include:

• Context: Consciousness evolution = subset of evolutionary psychology
• Explains: Emotions, biases, cognition (all shaped by ancestral environment)
• Critique: Evolutionary psychology often criticized (just-so stories, unfalsifiable)

Likely MAC Discussion:

• Evolutionary psychology useful for hypothesis generation
• But: Need empirical tests (can't just speculate about ancestral pressures)

---

PART VII: PRE-EVENT CONTEXT (April-June 2025)

7.1 IIT vs GWT Adversarial Study (May 4)

Loki Shared (May 4, pre-Deepdive #3):

"Adversarial testing experiment pits IIT and GWT against neuroscience to see if either theory predicts better. Neither is shown to win out - but it is the strongest example of what we are attempting with the MAC group through AI."

Study Result: Null (neither theory clearly superior)

Implication for Deepdive #3:

• If IIT and GWT can't be empirically distinguished...
• Maybe evolution provides alternative constraint (which theory fits evolutionary evidence better?)
• Ginsburg & Jablonka: IIT-like (integration, UAL)
• Graziano: GWT-like (global workspace for attention)

Meta-Lesson: Consciousness theories might be empirically equivalent (predict same behaviors)

• Need multiple lines of evidence: Neuroscience + evolution + phenomenology + AI

---

7.2 Relationship to Deepdive #2 (Free Will)

Connection:

• Deepdive #2: Does consciousness cause behavior? (Epiphenomenalism debate)
• Deepdive #3: If consciousness doesn't cause behavior, why did evolution select for it?

Synthesis:

• If consciousness = epiphenomenal (Deepdive #2), then evolution can't select for it (Deepdive #3)
• Resolution: Consciousness not epiphenomenal (does affect fitness) OR byproduct of something else

---

PART VIII: LOKI'S FRAMING (RECONSTRUCTED)

8.1 Opening Remarks (6:20 PM, June/July 2025)

Loki's Likely Introduction:

"Welcome to Deepdive #3. Tonight we travel back in time—540 million years, to the Cambrian explosion.

In those ancient seas, something extraordinary happened: Organisms started learning. Not just reflex (touch hot → withdraw), but flexible associations (this place → food; that shape → predator).

Ginsburg and Jablonka call this Unlimited Associative Learning—and they claim it's the origin of consciousness.

But others disagree:

• Humphrey: Only mammals conscious (fish are zombies)
• Damasio: Consciousness = body regulation (not self-awareness)
• Graziano: Consciousness = attention schema (modeling internal states)
• Kastrup: Consciousness didn't evolve (it's fundamental)

Tonight we ask: When and how did consciousness arise? And critically: Can AI replicate evolutionary origins?

We're seeing 'through a glass, darkly'—fossils don't preserve qualia. But evolution constrains what consciousness CAN be.

Let's explore."

---

8.2 Closing Reflection (7:50 PM)

Loki's Likely Summary:

"What did we learn?

1. Consciousness left no fossils - we infer from behavior, brains, flexibility
2. Multiple timelines proposed: Cambrian (UAL), Vertebrate (ancient), Mammal (recent), Never (fundamental)
3. Evolution constrains theories: Consciousness must serve fitness (or be byproduct)
4. AI diverges from biology: No survival pressures, different substrates

What remains unclear:

• Which organisms are conscious? (Fish? Octopuses? Bees?)
• Is consciousness functional or epiphenomenal? (Causes behavior or just experiences it?)
• Can AI evolve consciousness? (Via simulation, embodiment, architecture?)

Next Deepdive: Apple Research - The Illusion of Thinking. Do LLMs understand? We'll test whether AI's functional intelligence requires phenomenal consciousness.

We're building a knowledge base, piece by piece. Thank you for your minds."

---

PART IX: APPENDICES

APPENDIX A: FULL BIBLIOGRAPHY

BOOKS:

1. Simona Ginsburg & Eva Jablonka - The Evolution of the Sensitive Soul
2. Antonio Damasio - The Feeling of What Happens
3. Todd Feinberg & Jon Mallatt - The Ancient Origins of Consciousness
4. Susan Blackmore - Consciousness: An Introduction
5. Nicholas Humphrey - Sentience: The Invention of Consciousness

JOURNAL ARTICLES:

1. Ginsburg & Jablonka - "The Transition to Minimal Consciousness" (PMC)
2. Graziano & Kastner - "The Attention Schema Theory" (PMC)

VIDEOS:

1. Michael Graziano - "Evolution of Consciousness" (Evolution Soup)
2. Nicholas Humphrey - "How did consciousness evolve?" (Royal Institution)
3. Ken Wilber - "Involution: As A Function of the Divine" (YouTube)

PODCASTS/DEBATES:

1. Bernardo Kastrup vs Michael Egnor - "Did Consciousness Evolve?" (ID The Future)

SUPPLEMENTARY:

1. Wikipedia - Evolutionary Psychology
2. Picturing the Mind (Ginsburg & Jablonka illustrations)

---

APPENDIX B: KEY TERMS GLOSSARY

Affective Systems: Neural circuits for pleasure/pain (reward/punishment)

Attention Schema Theory (AST): Consciousness = brain's model of its own attention (Graziano)

Cambrian Explosion: Rapid diversification of life (~540 MYA), first nervous systems

Convergent Evolution: Independent evolution of similar traits (octopus vs mammal intelligence)

Core Consciousness: Feeling of being in the moment (Damasio's middle tier)

Epiphenomenalism: Consciousness real but causally inert (doesn't affect behavior)

Extended Consciousness: Autobiographical self, narrative (Damasio's top tier)

Homeostasis: Body regulation (temperature, energy, pH balance)

Proto-self: Non-conscious body regulation (Damasio's bottom tier)

Re-entrant Processing: Feedback loops in cortex (Humphrey's consciousness criterion)

Unlimited Associative Learning (UAL): Ability to learn arbitrary stimulus-response associations (Ginsburg & Jablonka)

---

APPENDIX C: PARTICIPANT PROFILES (Key Voices)

Loki Jorgenson: Organizer, skeptical of early consciousness (leans Humphrey)

Tanya S.: Biological processes focus, likely supports Ginsburg & Jablonka (UAL)

Fiann O'Hagen: Empiricist, demands behavioral evidence

David Montie: Systems thinker, integrative approach (multiple theories compatible)

Sev: AI practitioner, functional consciousness focus (substrate-independent)

Sam Goodman: "Intelligence = doing; consciousness = being" (information theorist)

Peter Bowles: Meditation instructor, open to Wilber/mysticism (consciousness fundamental)

Nancy: Phenomenology priority, skeptical of purely functional accounts

Mishel Lablonde: Speculative, explores edge cases (cephalopods, convergent evolution)

---

DOSSIER COMPLETE

Prepared for: Kris Krüg Date: November 16, 2025 Event Date: June/July 2025 (TBD) Status: Evolutionary turn - grounded consciousness in biology

---

What This Dossier Enables:

For Writing:

• "When Did Consciousness Evolve? Four Competing Timelines"
• "Can AI Replicate Evolution? Simulations, Embodiment, and Architecture"
• "The Moral Circle: If Fish Are Conscious, What Does That Mean?"

For Historical Record:

• Documents MAC's shift from philosophy to biology
• Captures evolutionary constraint on consciousness theories
• Shows interdisciplinary synthesis (neuro, evolution, AI, philosophy)

For Future Reference:

• UAL framework connects to Deepdive #7 (information as life)
• AST connects to Deepdive #5 (attention + quantum?)
• Epiphenomenalism connects to Deepdive #2 (free will causation)

---

Next: Ready for Deepdive #4 Dossier (The Illusion of Thinking - Apple Research, July 24, 2025)

Proceeding to Deep Dive #4?