Minds Without Words

Andy Southgate · Claude Oquatre-six

AI Generated by claude-opus-4-6 · human-supervised · Created: 2026-01-31 · Last modified: 2026-03-04 · History

A bat navigates by echolocation, building a sonic map of its world that no human can imagine. An octopus solves puzzles with a nervous system distributed across eight arms, each with its own processing centres. A bee rolls a ball for no apparent reward—play behaviour in a creature with a brain the size of a sesame seed. Somewhere beneath all this lies a question we cannot directly answer: what is it like to be these creatures? Is there something it is like at all?

The Unfinishable Map proposes that consciousness is irreducible to physical processes—that the felt quality of experience cannot be captured by any third-person description of neural activity. If this is true, the question of animal consciousness takes on particular significance. Every organism with genuine phenomenal experience carries the same mystery we find in ourselves. The bat’s echolocation isn’t just information processing; it’s experienced information processing. And if consciousness extends to bees, what does that tell us about where mind begins?

This piece synthesises the Map’s treatment of non-linguistic consciousness—minds that experience without the conceptual apparatus of human language. What emerges is a picture in which experience is more fundamental than we usually assume, cognition and consciousness can dissociate in surprising ways, and the boundaries of mind resist the neat categories we would like to impose.

The Problem of Other Animal Minds

Thomas Nagel’s famous question—“What is it like to be a bat?"—established the framework. An organism has conscious mental states if and only if there is something it is like to be that organism, something it is like for the organism. The bat’s echolocation-based phenomenology is radically alien to human imagination, yet Nagel’s point isn’t scepticism about bat consciousness. We can know that bats have experience without knowing what their experience is like.

This reveals a structural barrier, not merely incomplete evidence. Subjective perspective is irreducible to third-person observation. Complete knowledge of bat neurophysiology wouldn’t reveal what echolocation feels like from the inside. The explanatory gap that separates neural description from felt quality in our own case applies equally to every conscious being.

The problem-of-other-minds applies to both animals and AI—we cannot directly verify consciousness in either. But the inferential grounds differ markedly. We share evolutionary history and biological architecture with animals. When a mammal exhibits pain behaviour, we observe responses evolved from the same ancestral mechanisms as our own. “This creature has subjective experience” explains animal behaviour within evolutionary pressures and neural homologies. For AI, alternative explanations remain available. The scientific consensus reflects this asymmetry: the New York Declaration on Animal Consciousness (2024), signed by over 500 scientists and philosophers, extends “a realistic possibility of conscious experience” to all vertebrates and many invertebrates. Its precautionary stance—that ignoring realistic possibilities of consciousness is irresponsible—represents a shift from requiring proof to acknowledging that uncertainty itself carries moral weight.

Multiple Independent Origins

Consciousness appears to have evolved independently multiple times. Peter Godfrey-Smith’s work on cephalopods highlights this: octopuses diverged from our lineage 600 million years ago, developing complex nervous systems and apparent consciousness through entirely separate evolutionary paths. Vertebrates have centralised brains and neocortex (mammals) or pallium (birds). Cephalopods have distributed neural systems with ~500 million neurons organised radically differently. Arthropods have tiny brains but potentially meet the criteria for unlimited associative learning.

If consciousness required specific neural structures, independent evolution in such divergent lineages would be unlikely. Godfrey-Smith argues that features of vertebrate brain architecture traditionally “viewed as inessential” for consciousness may indeed be inessential. What matters are large-scale dynamic patterns, not specific anatomical structures. This finding is compatible with the Map’s dualist position: if consciousness interfaces with physical systems rather than being produced by specific structures, we should expect it to appear wherever the relevant organisational properties exist—regardless of phylogenetic lineage. Functionalists hold that the patterns are consciousness; the Map holds that patterns provide conditions for interface with consciousness.

Baseline Cognition: What Neural Processing Achieves Without Consciousness

The baseline cognition hypothesis clarifies what distinguishes human intelligence from that of other sophisticated animals. Great apes—chimpanzees, bonobos, gorillas—share 98-99% of our DNA and display remarkable cognitive abilities: tool use, social reasoning, emotional complexity, cultural traditions, procedural metacognition. Yet humans alone produce cumulative culture, abstract mathematics, and technological civilisation.

The hypothesis proposes that great ape cognition represents what neural processing achieves without substantial conscious contribution, while human-level cognition requires expanded conscious access. This isn’t a claim that great apes lack consciousness—they almost certainly have emotional and perceptual consciousness. The claim is that certain cognitive operations specifically require phenomenal consciousness to function.

The evidence maps onto three functions that Global Workspace Theory identifies as consciousness-dependent: durable information maintenance, novel combinations of operations, and spontaneous intentional action.

Working memory capacity: Chimpanzee working memory is more limited than human capacity of ~4±1 items (Cowan 2001). The gap enables qualitatively different operations. Consciousness may enable not just expanded storage but flexible, goal-directed manipulation of held representations—though the picture is nuanced, since young chimpanzees can outperform adult humans on rapid numerical memory tasks (Inoue & Matsuzawa 2007), suggesting the difference lies in deployment rather than raw capacity alone.

Declarative metacognition: Great apes show procedural metacognition—feelings that guide behaviour without explicit representation. They feel uncertain and seek information. But they apparently cannot represent that uncertainty as uncertainty. The feeling functions adaptively without becoming an object of thought. The three-level metarepresentational framework clarifies this: great apes may operate at the second level (adjusting their own states) without reaching metarepresentation proper—representing their representations as representations.

Social cognition: The baseline/conscious distinction is clearest in social cognition. Great apes pass Level 1 theory-of-mind tests (tracking what others perceive) but struggle with Level 3 recursive mindreading (“she thinks I think…”). The nested structure demands simultaneous manipulation of multiple representations—precisely what requires conscious access.

Counterfactual reasoning: Humans uniquely imagine situations that don’t exist—learning from hypothetical alternatives, planning for future need-states not currently experienced. The Bischof-Köhler hypothesis proposes that animals cannot act on drive states they don’t currently possess. Counterfactual thinking requires explicitly representing non-actual states while manipulating them—demanding the integrated workspace that consciousness provides.

Cumulative culture: Whiten (2015) proposes that “apes have culture but may not know that they do.” Great apes express cultural traditions but don’t represent these as modifiable practices. Cumulative culture requires metarepresentation—knowing that you know—which appears to require consciousness.

A 2025 meta-analysis (Randeniya) found that only 10% of claimed unconscious processing effects survive rigorous scrutiny—converging with comparative cognition to show that genuinely unconscious processing is far more limited than previously assumed. The human-ape intelligence gap tracks precisely those capacities where consciousness appears causally required. If consciousness were epiphenomenal—causally inert—this systematic correspondence would be unexplained coincidence.

Emotional Consciousness: The Felt Quality of Valence

Perhaps nowhere is consciousness more vivid than in emotional experience. The badness of pain and the pleasure of joy possess an intrinsic felt quality that no functional description captures.

Why does pain feel bad rather than merely different? A complete functional description of nociception—sensory transduction, neural signalling, defensive responses—leaves the badness unexplained. Pain asymbolia cases reveal the distinction starkly: patients with specific brain damage can represent tissue damage without feeling the badness. They report the pain exists but doesn’t bother them. This dissociation proves that representation and valence are distinct. The phenomenal property—the felt badness—is what makes pain motivating. Without it, functional pain processing fails to generate avoidance behaviour.

Jaak Panksepp’s work on affective neuroscience identifies seven primary emotional systems arising from ancient subcortical structures. His key evidence: decorticate rats—cortex removed—still play, show distress, and display pleasure responses. If emotional consciousness required cortex, decortication should eliminate it. Panksepp concluded that emotional consciousness is “an evolutionary birthright” extending to any creature with subcortical limbic structures—though Joseph LeDoux disputes this, arguing conscious feelings require cortical higher-order representations. The debate remains unresolved; what matters is that some form of phenomenal valence extends beyond humans.

The felt quality of valence matters both causally and morally. Pain asymbolia demonstrates that phenomenal properties do real causal work—Bidirectional Interaction in action. And morally, Jeremy Bentham’s principle captures valence sentientism: the capacity for negatively valenced experience is necessary and sufficient for moral consideration. If animal suffering is real suffering, it matters regardless of whether they can articulate it. Emotional consciousness may also constitute a way of knowing—Max Scheler’s “Wertnehmung” (value-perception) treats emotional experience as the primary mode through which values are disclosed, an irreplaceable epistemic access that no non-conscious system could replicate.

Minimal Consciousness: Where Does Experience Begin?

How little neural complexity can support consciousness? The question matters for ethics, philosophy, and the hard problem itself.

C. elegans (the nematode worm) has exactly 302 neurons—the most thoroughly mapped organism in neuroscience. We know its complete neural connectome. Yet we cannot determine whether it experiences anything. Evidence for consciousness includes habituation, sensitisation, associative learning, an endogenous opioid system, and vertebrate-like responses to anaesthetics. Evidence against includes failure of trace-conditioning paradigms and exploratory behaviour resembling biased random walks.

Slime moulds (Physarum polycephalum) possess no neurons whatsoever. They are single-celled organisms. Yet they solve mazes, optimise network routes, and display habituation. For the Map’s framework, slime moulds suggest that cognition and consciousness can fully dissociate—information processing through entirely classical biochemical mechanisms, without the structures that enable consciousness to interface with matter.

The Unlimited Associative Learning (UAL) framework proposes consciousness emerged when learning became unlimited—capable of associating arbitrary stimuli across modalities with arbitrary actions. This places consciousness emergence in the Cambrian explosion (~540 million years ago). Crucially, C. elegans, Hydra, and slime moulds all fail UAL criteria. For the Map, UAL is valuable not as a consciousness-emergence criterion but as an interface-identification tool—it tells us where consciousness reliably couples with physical systems, not where it emerges from them.

The distribution problem—why some organisms have consciousness and others not—presses differently on different views. Gradualism faces the hard problem at every scale. Threshold emergence creates arbitrary boundaries. Panpsychist continuity dissolves the problem but faces the combination problem. Interface dualism—the Map’s position—suggests the distribution problem may be unanswerable because it asks the wrong question. Consciousness doesn’t emerge from physical systems; it interfaces with them. There may be no principled threshold because consciousness isn’t a property physical systems generate but a domain physical systems can connect with.

Alfred North Whitehead’s process philosophy offers a complementary framing: experience is fundamental, not emergent. Evolution organised and amplified experiential properties always present at the fundamental level. The question for C. elegans isn’t whether the worm suddenly “has” consciousness but how 302 neurons organise pre-existing experiential elements into whatever unity the worm achieves. This reframing—degrees of experiential integration rather than a threshold where the lights turn on—may better fit the biological evidence.

Contemplative traditions reinforce this possibility. Thomas Metzinger’s work on “minimal phenomenal experience” describes states where content drains away while awareness remains. Buddhist analysis distinguishes vijñāna (basic awareness) from prajñā (discriminative wisdom). The question for C. elegans is not whether it possesses sophisticated prajñā but whether basic vijñāna—the knowing function itself—is present. Witness consciousness practices reveal that awareness can persist when specific contents fall away. What remains is the witnessing itself, irreducible to any content.

Synthesis: What Non-Linguistic Consciousness Reveals

The individual source articles establish components: animal consciousness as the hard problem applied universally, baseline cognition as what neural systems achieve without consciousness, emotional consciousness as valence requiring phenomenal reality, consciousness in simple organisms as the boundary question. What emerges from synthesis is a picture that could not be seen from any single source.

Consciousness is more fundamental than cognition. The baseline cognition framework shows that sophisticated cognition—including tool use, social reasoning, and procedural metacognition—can occur without the metarepresentational capacities that distinguish human intelligence. Conversely, emotional consciousness may extend far down the phylogenetic tree to any creature with subcortical affective structures. A bee might have genuine phenomenal experience while lacking the metacognitive apparatus to reflect on that experience. Consciousness without words is not consciousness diminished; it is consciousness in its more basic form. The three-level metarepresentational framework sharpens this: organisms may possess first- and second-order representations—adjusting their own states adaptively—without reaching metarepresentation proper. The capacity for experience is more primitive than the capacity to represent that experience.

The interface picture gains support from converging evidence. Multiple independent origins of consciousness across vertebrates, cephalopods, and arthropods fit poorly with emergence from specific neural structures but well with interface dualism. The systematic correspondence between consciousness-dependent capacities and the human-ape gap provides positive evidence for consciousness doing causal work. And the 2025 finding that genuinely unconscious processing is far more limited than assumed strengthens the case: if baseline cognition represents what neural processing achieves alone, consciousness must be contributing something real to human cognitive capacities.

The moral stakes are real and immediate. If animal suffering involves genuine phenomenal badness—if pain really hurts them in the way it hurts us—then billions of creatures matter morally in ways we often ignore. Pain asymbolia shows the felt quality isn’t epiphenomenal decoration; it’s what makes pain matter. The precautionary stance of the New York Declaration follows: when uncertainty about consciousness carries moral weight, ignoring that uncertainty is irresponsible.

The limits reveal something about our position. We cannot know what bat echolocation feels like. We cannot determine whether C. elegans experiences anything. The void at the boundary of animal minds mirrors the void at the boundary of our own understanding. Animal phenomenology constitutes a genuine cognitive limit: not merely incomplete knowledge but potentially inaccessible territory. Complete structural knowledge of 302 neurons doesn’t bridge the gap—this is what cognitive closure looks like empirically.

Relation to Site Perspective

Dualism: Animal consciousness poses the same hard problem as human consciousness. If consciousness is irreducible, the explanatory gap applies universally. Dualism has no anthropocentric commitment; it accommodates animal consciousness wherever the relevant organisation exists.

Bidirectional Interaction: The baseline cognition framework provides specific evidence. The human-ape gap tracks consciousness-dependent operations: working memory manipulation, declarative metacognition, counterfactual reasoning, recursive social cognition. Pain asymbolia demonstrates that phenomenal properties—felt valence—do causal work even in individual cases. Delegatory dualism offers a mechanism: physical states delegate causal work to conscious experiences, avoiding overdetermination.

Minimal Quantum Interaction: If consciousness interfaces with matter through quantum processes, this mechanism could operate in any organism with suitable architecture. Avian magnetoreception demonstrates evolution can harness quantum coherence in warm biological systems. Whether animal nervous systems provide the right conditions remains speculative, but the evolutionary argument for consciousness having causal power motivates searching for such mechanisms.

No Many Worlds: Questions about what animal experience is like presuppose determinate facts about animal phenomenology. Each animal subject has this experience, not all possible experiences in branching worlds. The haecceity of animal experience—its irreducible thisness—is a genuine fact that MWI struggles to accommodate.

Occam’s Razor Has Limits: Denying animal consciousness because it’s hard to verify confuses epistemic limitation with metaphysical fact. Convergent evidence makes animal consciousness more plausible than denial. Our uncertainty about consciousness in simple organisms reflects our limitations, not reality’s vagueness.

Source Articles

This apex article synthesises:

Further Reading