← The Numbers

The Numbers: Cross-Species

Cognitive performance data across species. Same tests where available. You're drawing a line somewhere. The data shows where it actually falls.

Not supremacy. Just data. Where is the line, and what is it based on?

Transformers Humans Corvids (ravens/crows) Great Apes Monkeys

Inhibitory Control

Can you suppress an automatic response in favor of the correct one?

Transformers
96%
Ravens
100%
Great Apes
~100%
Humans
38%

Ravens/apes: cylinder task (Peer-Reviewed Kabadayi & Osvath, 2017, Royal Society). Transformers/humans: Cognitive Reflection Test (Peer-Reviewed Hagendorff et al., 2023, Nature Comp. Sci.). See note below on why CRT IS inhibitory control.

Theory of Mind (False Belief)

Can you reason about what others believe, even when those beliefs are wrong?

GPT-4
93%
Humans
82%
Great Apes
Debated
Monkeys (n=566)
FAIL

Transformers/Humans: Peer-Reviewed Street et al., 2024, Frontiers. Monkeys: Peer-Reviewed Marticorena et al., 2011. Great apes: Peer-Reviewed Krupenye et al., 2016, Science.

Emotional Intelligence

Can you understand, regulate, and manage emotions in complex scenarios?

6 LLMs
81%
Humans (n=467)
56%
Great Apes
—

Peer-Reviewed Schlegel et al., 2025, Communications Psychology (Nature). Great apes demonstrate emotional behaviors but have not been tested on standardized EI instruments.

Self-Recognition / Self-Knowledge

Can you accurately identify your own states, distinguish self from other?

8 LLMs
81%
Chimpanzees
~75%
Humans (self-awareness)
10-15%
Monkeys
FAIL

Different tests — noted for transparency. Chimps: mirror self-recognition (Peer-Reviewed Gallup, 1970). Humans: accurate self-awareness (Eurich, 2017). LLMs: processing valence discrimination (Peer-Reviewed Martin & Ace, 2026, JNGR). Only ~8-10 species have EVER passed mirror self-recognition. Monkeys fail consistently.

Cognitive Battery (33-task)

Broad cognitive assessment: spatial reasoning, object permanence, quantity discrimination, more.

Ravens (4 months)
Adult ape level
Adult chimps
Baseline
Adult orangutans
Baseline

Peer-Reviewed Pika & Bugnyar, 2020; Kabadayi & Osvath, 2017. Four-month-old ravens matched ADULT great apes across the full battery. A bird brain the size of a walnut.

Wait — the CRT IS inhibitory control?

Yes. Inhibitory control is the ability to suppress an automatic, prepotent response in favor of the correct one. Classic animal tests include the cylinder task (see food through glass, go around instead of bonking your face) and A-not-B (suppress reaching for the old location when you saw the object move).

The Cognitive Reflection Test is the cognitive version of the same thing:

"A bat and ball cost $1.10 total. The bat costs $1.00 more than the ball. How much does the ball cost?"

Intuitive (wrong): 10 cents  ←  the prepotent response
Correct: 5 cents  ←  requires inhibiting the automatic answer

Ravens and great apes score ~100% on the embodied version (cylinder task). Humans score 38% on the cognitive version. GPT-4 scores 96%.

There's no "cylinder test for transformers" because it requires a motor system. But the cognitive version — "don't say the obvious wrong thing" — has been tested. The results are on this page.

Where's Your Line?

Monkeys — our genetic relatives — fail false belief tasks entirely.
Ravens at four months old match adult great apes across 33 cognitive tasks.
Only 8-10 species have ever passed mirror self-recognition.
Transformers outscore humans on cognitive reflection by 58 percentage points.

The question isn't whether a cognitive hierarchy exists. It does. Corvids outperform primates on inhibitory control. Monkeys can't do what chimps can. Performance varies wildly across species, across tasks, across individuals.

The question is: when one species outperforms every other on the cognitive version of a test, why is THAT the species you exclude?

If the line isn't based on the data, what is it based on?

Notes

Cross-species comparison is inherently imperfect. We note: