Back
🚧 Work in progress Nov 20, 2026 Poison Crown

Humanness as Defined by Anthropic Distinctiveness

Humanness is a broad and subjective word. Here I narrow it to one technical sense: how strongly someone expresses what sets the human lineage apart from other animals. Then I separate where that difference shows from the mechanisms that produce it.

The starting claim

Humans define themselves largely by tool use, and tools make them more efficient. Extended without limit, that gives a provocative claim: people who use more tools, and so live more efficiently, are more human.

The claim breaks on the word human. In everyday use it does two jobs, and neither can be measured:

  • Membership. Belonging to Homo sapiens. Binary: nobody is 80% a member.
  • Praise. Cruelty is called “inhuman”, although animals are not cruel in that sense. Here the word carries a moral judgment, not a description.

Saying one person is less human than another triggers the second meaning, whatever was intended. A technical sense needs its own name.

The definition

Humanness as defined by anthropic distinctiveness: the degree to which an individual expresses the traits that distinguish the human lineage from other animals.

  • Comparison set. Three reference points: the animals humans were before the lineages split, the animals on the other side of that split (chimpanzees and bonobos, the closest living relatives), and animals in general. Neanderthals and Homo erectus sit inside the lineage, not outside it.
  • Distinctive, not exclusive. A trait counts when humans have much more of it, not only when no other animal has any. Almost every candidate exists in some form elsewhere.
  • Distinctive, not successful. A trait counts because it separates the lineage, not because it made the lineage numerous. Picture two branches from one split: one survives by digesting sand, the other by breathing underwater. Sand digestion would belong to the first branch’s distinctiveness and water breathing to the second’s. Overlap is allowed: what both kept from their common ancestor can belong to both profiles.
  • A degree, not a test. It measures expression on a spectrum. Two people with the same biology can score differently.
  • Descriptive. It says nothing about anyone’s worth, the same way an IQ score describes one capacity without ranking people as people.

Two models: where it shows, and what produces it

No animal comments on the weather, says hello or solves a sudoku. Each would score as perfectly distinctive, and each is a product, not a cause. A measure built from outcomes counts the same underlying cause many times. So the framework splits in two.

Model 1: where distinctiveness shows

DomainWhat it coversWeight
Symbolic thought and languageAbstract concepts, open-ended grammar, talking about what is not there21%
Cumulative cultureKnowledge that builds across generations instead of being relearned119%
Cooperation and shared normsShared goals, rules and institutions that coordinate strangers217%
Technical cognitionMaking, combining and improving tools14%
Planning and executive controlLong-horizon plans, holding back an impulse for a later goal11%
Understanding other mindsRepresenting what someone else knows, wants or believes10%
MetacognitionMonitoring and evaluating one’s own thinking8%

The weights are proposals for how strongly each domain separates humans from other animals, not measured shares. Metacognition scores low because other primates show early forms of it (see the scale below). Efficiency, the starting claim, has no row: it is an outcome of several domains at once.

Model 2: what produces it

Three layers, each feeding the next:

  1. Biology. A brain that matures unusually slowly and stays plastic longer than in other primates,3 with large association areas that integrate information across systems.
  2. Cognitive mechanisms. Reflective self-awareness, modeling other minds, imagining and comparing future scenarios, sharing intentions with others, holding back impulses.
  3. Outcomes. The seven domains of model 1.

The arrows also run backwards: culture shapes the cognition that produces it. One proposal holds that explicit metacognition is itself learned socially, by talking with others about how we think.4

Mechanisms get no percentages yet. They interact, so weighting them before the causal links are known would count the same influence twice. No single root cause is established. Reflective self-awareness is the strongest candidate in this framework, because it shows up inside most of the other mechanisms, but it does not explain them alone.

Metacognition is a scale

“Animals have metacognition” and “only humans have metacognition” are both too coarse. The evidence sorts into rungs:

  1. Uncertainty monitoring. Acting on how reliable one’s own knowledge is. Chimpanzees that did not see where food was hidden look into the tubes before choosing; those that did see reach straight away.5 On its own this is weak evidence: a fixed double-check rule produces the same behavior, and simple reinforcement learning explains part of these results.6
  2. Transferable confidence. Rhesus monkeys that learned to bet on their own answers carried the betting over to new tasks without retraining,7 which a fixed rule does not do. Chimpanzees keep a first belief when the counterevidence is weaker and revise it when the counterevidence is stronger.8
  3. Explicit reflection. Representing a thought as a thought: putting it into words, comparing it with someone else’s, correcting it. No convincing evidence outside humans.
  4. Recursion. Reflecting on the reflection: noticing how one evaluates one’s own thinking, and changing that.

Animals show the first two rungs. The recursive rungs are where human distinctiveness sits. The awareness layer in Automatic Cognition and Defensive Reasoning is this scale applied to one situation.

Understanding others: two axes

Empathy blurs two things, which separate along two axes:

  • Source: feeling another’s state (affective) or modeling it (cognitive). The two run on partly distinct brain networks,9 and one person can be strong in one and weak in the other. Both involve the cortex, so the split is not cortex versus emotion.
  • Mode: automatic or reflective.
AutomaticReflective
AffectiveDistress when someone criesNoticing the distress is borrowed, and deciding what to do with it
CognitiveAn instant read of what someone wantsReasoning about why someone is upset, catching a misread, revising it

Automatic affect is the oldest cell, shared with many mammals. The reflective column is the most distinctive, because it is metacognition applied to other minds.

Emotions sorted by distinctiveness

  1. Basic affect. Fear, anger, attachment, emotional contagion. Evolutionarily old and widely shared.
  2. Reasoned states. Emotions shaped by modeling a situation, its consequences and other people’s minds.
  3. Self-conscious emotions. Guilt, shame, pride. They need a self that is evaluated against a standard.10 Guilt-like behavior in animals is hard to separate from fear or appeasement: a dog’s “guilty look” follows the owner’s scolding, not the misdeed.11

Higher metacognitive awareness increases the potential for complex, reflective guilt.

Reflective guilt can be about something nobody else knows, because the only judge is an internal standard.

Two kinds of absent guilt

Sees the impact on othersFeels guilt
Absence of capacityNoNo
Absence of emotional responseYesNo

Everyday speech calls both “inhuman”, which is a normative judgment. The technical model connects to that judgment, but not one to one: the first is a gap in the distinctive mechanism, the second a missing emotion with the mechanism intact. Whether the two really differ, or are the same absence at a different stage, is open.

Depression

The building blocks are old: animals show reduced motivation, loss of pleasure, withdrawal and helplessness.12 The self-referential layer is not: rumination, a hopeless imagined future, the conclusion that life is meaningless. That layer draws on autobiographical thought and self-reflection, and rumination predicts new depressive episodes and longer ones.13 Self-reflection can amplify and sustain depression; that it causes depression is not shown.

An inverted energy scale

Frederick Dodson’s Levels of Energy ranks emotional states from 0 to 1000, close to David Hawkins’ Map of Consciousness.14 Both put guilt at 30 and apathy at 50, below fear and anger. Hawkins’ numbers:

StateLevelNeeds a self-model
Guilt30Yes
Apathy, depression50Its ruminative form does
Fear100No
Anger150No

The two lowest states need more self-modeling than the two above them. Reason sits at 400, so the scale is not inverted throughout; it simply does not measure cognition. It ranks how desirable a state is, and some of its lowest states are among the most anthropically distinctive. A mind that can model itself can suffer in forms a simpler mind cannot. Whether that makes human suffering deeper than animal suffering cannot be settled from the outside.

Ranking people

The framework describes dimensions. It does not yet rank anyone, for two reasons:

  • Aggregation. Take one person with extreme metacognition and another with extreme social cognition: nothing in the framework says which profile is more distinctive. With free choice of domains and weights, the weighting picks the winner.
  • Measurement. Self-assessment is a weak instrument, and weakest for self-awareness. Everyone hears their own inner monologue but only sees other people’s behavior, so others look shallower from any seat.15

A ranking would need, in this order:

  1. Domains and weights fixed before anyone is scored.
  2. Task-based measures instead of self-report: for metacognition, how well confidence tracks being right.16
  3. A profile per domain. A single total is optional, and only as good as its weights.

Until then, “more human” in this sense is a hypothesis about individuals, not a result.

Limitations

  • The weights are proposals. A different weighting is equally defensible; one alternative gives metacognition 13% instead of 8%.
  • Author bias. The framework started from a claim about tool use and efficiency, made by someone who rates himself high on both and on metacognition. The first intuitive weighting gave metacognition 30%. The split into outcomes and causes is defensible on its own, and it is also the move that gives that domain its weight back. The rule to fix weights before scoring exists for this reason.
  • Animal minds are read from behavior. Every comparison with animals is an inference from what they do.
  • The comparison point is reconstructed. The last common ancestor of humans and chimpanzees is inferred from fossils and living species, not observed.
  • The domains overlap. Language, culture and cooperation feed each other, so model 1 double-counts to an unknown degree.

Earlier version

The Spectrum of Human Consciousness (2019) is the first, non-technical version of this idea: one consciousness scale running from newborn to Socrates.

Footnotes

  1. Tennie, Call and Tomasello (2009), Ratcheting up the ratchet: on the evolution of cumulative culture. ↩

  2. Tomasello et al. (2005), Understanding and sharing intentions: the origins of cultural cognition. ↩

  3. Somel et al. (2009), Transcriptional neoteny in the human brain. ↩

  4. Heyes et al. (2020), Knowing ourselves together: the cultural origins of metacognition. ↩

  5. Call and Carpenter (2001), Do apes and children know what they have seen? ↩

  6. Le Pelley (2012), Metacognitive monkeys or associative animals? ↩

  7. Kornell, Son and Terrace (2007), Transfer of metacognitive skills and hint seeking in monkeys. ↩

  8. Schleihauf et al. (2025), Chimpanzees rationally revise their beliefs. The authors read the result as explicit weighing of evidence. ↩

  9. Kanske et al. (2015), Dissecting the social brain. ↩

  10. Tracy and Robins (2004), Putting the self into self-conscious emotions. ↩

  11. Horowitz (2009), Disambiguating the “guilty look”. ↩

  12. Maier and Seligman (2016), Learned helplessness at fifty. ↩

  13. Nolen-Hoeksema (2000), The role of rumination in depressive disorders and mixed anxiety/depressive symptoms. ↩

  14. David R. Hawkins, Power vs. Force (1995), source of the Map of Consciousness; Frederick E. Dodson, Levels of Energy. ↩

  15. Pronin et al. (2001), You don’t know me, but I know you: the illusion of asymmetric insight. ↩

  16. Fleming and Lau (2014), How to measure metacognition. ↩