About a month ago, one of my regular respondents asked me to blog about self-awareness in animals. I’m doing so now because it will be useful for an essay I’m planning to write about ethical and legal definitions of humanity.
Let’s start by defining a slightly more abstract category, what science-fiction fans call a ‘sophont’. A being or animal is fully sophont if it:
- Shows behavioral evidence of self-awareness, e.g knows it has a mind.
- Shows awareness of the existence of other minds.
- Communicates abstractions using language.
- Invents and uses tools.
- Can reason about counterfactual hypotheticals, e.g. can answer questions of the form “What would things be like if a fact I am now observing to be true were not true?”
(That last one may seem odd to those of you not familiar with primate ethology, but bear with me…)
Normal human beings pass all these tests. Few animals are known to even come close to passing all of them — but the “are known” is necessary, because in some cases animals might pass all of the tests other than tool use without being able to communicate that to humans.
Some people feel quite strongly about animal rights. Their position does not, however, generally spring from an assertion that animals are sophonts; rather, it seems to come down to an unwillingness, or inability, to distinguish between sophont status and sentience — that is, being able to feel. In practice, while theoretically even insects can feel, even PETA members tend to ascribe sentience only to animals that can exchange recognizable emotional signals with us — which is to say, basically, mammals.
The mamalian repertoire of behaviors for communicating states like fear, affection, anger, boredom, and playfulness is remarkably conservative. So much so that humans can have meaningful emotional communication with cats, dogs, and raccoons, a datum that would be astonishing if we weren’t so used to it! Some mammals are so good at this that we routinely keep them around for pleasure.
Even dogs and cats exhibit little evidence of sophont behavior, though. They can learn tricks by reinforcement, but they don’t use tools, they show only very weak problem-solving intelligence and even less evidence that they have a theory of mind or self-awareness. One test animal behaviorists use for self-awareness is whether the animal will try to remove a smudge from its face when it sees itself in the
mirrors; cats and dogs generally fail this. Sentience is less than sophont status.
There are, however, interesting borderline cases among the animals. Chimpanzees, bonobos, gorillas, orang-utans, dolphins, whales, seals, otters, elephants, a few species of birds, and even octopi and squid have all displayed not merely sentience but suspiciously sophont-like behaviors. Here are some data:
- Great apes can learn sign language and use it in original ways, including coining new terms and making puns; they also make and use tools and have complex social patterns recognizably akin to human ones. Some apes can pass the mirror test. Some handle abstractions tolerably well, and there is a famous incident in which a gorilla named Koko speculated to its trainers about an afterlife after its pet kitten died.
- There is some, disputed, evidence that dolphins can in fact learn to use iconic languages. Nobody who has interacted with dolphins at any length is in any doubt that they have minds, recognize human beings as having minds, and have a lively sense of humor (not infrquently they try to include humans in their play). They can count and have at least a limited ability to handle abstractions. Claims that they are sophont at human level are not established by evidence, but nobody who has studied the evidence would be much surprised if
they turned out to be.
- Elephants use tools, engage in complex vocal communication (parts of which are infrasonic, entertainingly enough), have elaborate social behaviors, play jokes on their trainers, and apparently grieve for their dead. They pass the mirror test. And they really do have long memories; they’ve been known to wait years for a chance to take revenge an abusive human. New evidence suggests they may be very close to human in intelligence and self-awareness.
- Some parrots and crows can count. Some talking birds appear to graduate from mimicry to using learned words to express emotional states, and a few seem to make the jump to talking about the emotional states of humans.
- Mediterranean squid have a very high brain-mass-to-body-mass ratio (this correlates with various indicators of intelligence in mammals). They been observed to solve mechanical problems (how to extract the tasty food fish from the corked bottle) by examining their environment, thinking for a while, then acting quickly and correctly. Octopi and squid appear to communicate within groups using color flashes made with chromatophores in their skins; nobody know what they’re communicating, exactly, but the potential bandwidth of that channel is extremely high. There is even some evidence of observational learning in octopi.
Significantly, the birds who seem to be using speech, and the apes who handle abstractions best, are animals that have lived with humans for a long time. Even cats and dogs, though not very bright compared to elephants or dolphins or chimps, accasionally show flashes of self-awareness (for example, by recognizing themselves in a mirror).
Given all this animal data, what’s left as a unqiquely human capability? Though interpretation of the experimental results is controversial, one thing even the cleverest nonhuman primates seem to have problems with are counterfactual hypotheticals.
Imagine a table with two red balls, three green balls, and an upright paper screen large enough to hide a sixth ball behind it. To human beings, the following two questions are both easy and nearly indistinguishable:
- How many green balls would be in the room if there were a green ball behind the screen?
- How many green balls would you be able to see if the leftmost green ball were red?
But these two questions are subtly different. The first is a what-if that is consistent with all visible facts: the second is a what-if that contradicts a visible fact (thus ‘counterfactual’).
The brightest non-human primates handle questions like the first one pretty well, but questions like the second one rather poorly. They don’t seem to have have the capacity to construct a full-blown hypothetical universe in their heads and reason about it despite what observable reality is telling them.
(This test has only been done with primates. You need to have a language in common with your subject(s) to do it; primates can use sign language or symbol tiles, but communication with other possible quasi-sophonts is far more limited so far.)
If this test is to be believed, what distinguishes humans from other higher primates is not reasoning ability but imagination!