Has anyone ever proposed a Turing test for a non-human biological entity

June 17, 2018 2:32 AM

I'm just curious if anyone or if any group has proposed a way to breed a non-human species for sentient behavior? Sure, some dogs and parrots come close to being as intelligent as a "five year old" -- but I assume no one has really pushed the boundaries. Or am I wrong? Is this a mad-scientist ethical nightmare? Or could it be done in an ethical way?
posted by mhh5 to Science & Nature (18 answers total)

I think you misunderstand the nature of intelligence and sentience. Human understandings of animal sentience and cognitive abilities are not well developed and severely hindered by human hubris. This idea of "intelligent as a five year old" means what? Cognitive intelligence? Emotional intelligence? Physical intelligence? Maybe a species such as a rat or a dolphin or a whale or an elephant is, in its own context, already more intelligent than a human.

IMHO, it would be unethical to create a non-specific sentient being just to see if you could. What are you going to do with it later? Frankenstein's monster didn't fare so well. Do you want it to be constantly aware of its own singularity, loneliness and its subjugation to human demands? That doesn't seem very humane
posted by Thella at 3:44 AM on June 17, 2018 [2 favorites]

If by sentience you mean having conscious experience, my own belief is that many animals already have conscious experience. It’s hard for me to make sense of their behavior otherwise.
posted by eirias at 4:36 AM on June 17, 2018 [4 favorites]

All animals are sentient. Some are sentient in ways we humans recognize as close to our own way of feeling, thinking and perceiving. And these animals, in many ways, are often more intensely sentient than we are, able to feel and respond to the world in highly sensitive ways.
Watching urban deer or foxes also shows animals that seem more sentiently engaged with each other than humans. Sometime watch a doe wait for her fawn to join her across an invisible expanse, scanning for the fawn and keeping an eye out for threats, and then touch noses and run off together, to see the ways their interaction relies on senses of perception connected to bonding that either we don't have, or have in much less effective levels.
posted by nantucket at 4:38 AM on June 17, 2018 [3 favorites]

Dogs have been bred for intelligence for many years, for example sheep dogs.

The Turing test requires fluent human language and writing, so no non-human animals are currently close to that https://en.wikipedia.org/wiki/Turing_test

Some great apes have been taught simple sign language https://en.wikipedia.org/wiki/Great_ape_language. There is some discussion about possibilities of selectively breeding great apes in this Quora question: https://www.quora.com/Would-it-be-possible-to-breed-chimpanzees-as-intelligent-as-humans-if-one-mated-only-the-smartest-chimps-from-each-generation-for-many-years

I don't think there is anything within reach of current breeding techniques that could increase an animal's intelligence enough to let it pass a Turing Test, nor is it being attempted due to ethical concerns.

"Sirius" (Olaf Stapledon, 1944) is a sci-fi novel about a dog bred for intelligence that has language.
posted by richb at 5:44 AM on June 17, 2018 [2 favorites]

Best answer: I suspect that breeding for human-like intelligence would take a very long time.

There have been some examples of relatively quick directed evolution, but all of the ones that I'm familiar with are cases where an already-existing feature was kept on or turned off. A Soviet experiment bred foxes for domestication over the course of 50 years or so; what happened, as in most cases of domestication, was neoteny, in which juvenile features were retained into adulthood. You can evolve your own multicellular yeast over the course of a few weeks, but the mechanism is incomplete mother-daughter cell division. The Long Term Evolution Experiment got citrate utilization in E. coli after 30,000 generations, which was initially touted as the addition of a brand new feature by way of evolution, but it turns out that it was merely a case where "a gene had been turned on under conditions where it was normally turned off," and the same effect can be reliably achieved with directed evolution in 50-100 generations.

If human intelligence is the result of neoteny, then perhaps it could be evolved quickly (in 50 or 100 generations, say) in other species. I haven't encountered anyone who has suggested that, though, and I don't know of any domesticated animals which are smarter than their wild counterparts. (I haven't done any research on that, and am happy to be corrected if I'm wrong.) There are many hypotheses to explain the evolution of human intelligence with various degrees of empirical support; among the more prominent are the social brain hypothesis (PDF - big, complicated social groups require big, complicated brains) and the frugivore hypothesis (PDF - to live on fruit requires cleverness and great long-term memory).

It would be interesting to design a breeding environment which selected for both of these. How would you do it?

The problem, though: How many years would each generation take, if you're selecting for the animals which can keep track of and reason about complex social and environmental facts over multiple years? Let's say that your generations take 5 years each. That's optimistic, since intelligent animals like dolphins, elephants and chimps take at least 8 years to reach sexual maturity. (Brain size is correlated with age at sexual maturity; if you allow brain size as a rough stand-in for intelligence, this suggests that building smarter animals takes longer. As per that paper, humans are something of an anomaly: "If humans were typical primates, we would expect from the regression that humans would become sexually mature at about age 44.") The most optimistic situation is that the ability is already latent in the animal's genome - as in the cases of the domesticated foxes, snowflake yeast, and citrate-utilizing E. coli examples above - and you can bring out the already-existing intelligence in 50-100 generations. That's a 250-500 year experiment with 5-year generations.

The more likely case is that it would take many more generations than that. Putting aside the ethical issues, it's an impractical breeding experiment.

As a GMO experiment, though, perhaps you could overcome the ethical scruples of these researchers and inject human glial cells directly into monkey brains.
posted by clawsoon at 8:10 AM on June 17, 2018 [1 favorite]

Best answer: Putting aside the ethical issues, it's an impractical breeding experiment.

I'm quite surprised that such breeding-for-intelligence experiments have not been tried on small, fast-breeding and still remarkably intelligent animals like rats, which are already a common subject for the measurement of cognitive abilities (the "general intelligence" section of the Rat article on Wikipedia is a mess, unfortunately; I'm not sure what the consensus is). It would take about 8 years to breed 50 generations, and it would be (a little) less ethically problematic since one wouldn't expect to rats to reach primate-level intelligence anyway.
posted by elgilito at 9:00 AM on June 17, 2018

Best answer: Rat breeding experiments for intelligence have been done:

A 1940s psychologist named Robert Tryon wondered if rats could be bred to complete a maze more competently, and after seven generations of selective breeding he succeeded. ...

What he didn’t show was that intelligence can be bred. The rats weren’t smart. They were “maze bright.” They were bred to have one ability, and they had it. ...

Exactly what makes these rats bright or dull is still being studied. It helps that the maze bright and maze dull strains of rat have been maintained since the 1940s. They’re now called S1 and S3 and are still available for experiments.

This has given us some interesting takes on “maze bright” and “maze dull.” One study concluded that the strains didn’t differ in intelligence, or even ability, but motivation. The maze dull rats just didn’t care about what they were being trained to do. Another study, done by psychology students in 1963, found clear differences in learning ability. The maze bright strain kept improving over time, while the maze dull strain showed slight improvement early in their training, but then leveled off. It was only after the study was done that the students were informed that they’d actually been given a random group of rats, none of which were maze bright or maze dull.

If nothing else, it's an object lesson in the many pitfalls awaiting the intelligence researcher.
posted by clawsoon at 9:11 AM on June 17, 2018 [8 favorites]

Since you asked about ethics, I would ask what purpose would be met that would make this worth subjecting animals to these experiments. I think most people no longer believe it is OK to test animals for cosmetic purposes. Other animal research gets more complicated. Most people would probably say animal experiments are OK if the end result is curing human cancer, but what if the end result is making cold medicine last a little longer or extending the patent on a medication that already works? People who are concerned about animals used in medical research are often painted as idiots, but if you look at the issue in all of its complexity, there are some real problems, including the question of how often these experiments are really applicable to humans. I would ask you if it’s worth it to you to have a dog you care about used for this experiment.
posted by FencingGal at 9:28 AM on June 17, 2018

There's the Mirror test for self-awareness (or at least self-recognition).
posted by WasabiFlux at 9:59 AM on June 17, 2018

David Brin has a series of novels about this and he brings up many issues. Can't remember which one I read, but it was interesting.

Some whale and dolphin species appear to have complex intelligence and communication, as well as other primates in addition to humans. Octopii maybe, as well. Corvids (crows, ravens) have well-developed family and community, use tools, solve problems. Elephants. Many animals see better, and differently, than humans, have a better sense of smell, are stronger, etc. Standing upright allows humans to use their hands way more effectively, developing opposable thumbs that give tremendous dexterity and combined with standing upright, we developed tools and technology.

On a consistent basis, humans make and use complex tools, have self-awareness, use complex speech & language, have history. If this is a strong area of interest, read Origin of Species, or read about it. I'm a fan of Robert Sapolsky, strongly recommend Primate's Memoir. Anthropology and Primatology sections at your library.
posted by theora55 at 11:01 AM on June 17, 2018 [2 favorites]

Answering your jokey title: http://www.dailygalaxy.com/my_weblog/2011/01/future-tech-how-would-we-determine-the-level-of-extraterrestrial-artificial-intelligence.html

I dated one of the authors for a while and learned far more than I expected about Turing and intelligence tests.
posted by b33j at 3:34 PM on June 17, 2018

FencingGal, hmm. You ask what the value of this kind of experiment might be? Imagine if it's successful. Demonstrating that a Turing-test-passing intelligence can be created from a species that previously didn't have that ability would be a huge achievement. It would alter how humans think about ourselves in the universe -- analogous to shifting from a geocentric universe to a heliocentric one. It would put religions into deep(er) doubt. If there was a scientific method to engineer a more thoughtful brain using advanced genetics, would we discover that there is an upper limit to biological intelligence as well? Or would we learn ever more about biological intelligence and find a path to a "biological singularity" where brains could be grown at exponential rates on demand?

Testing animals in a cruel way to ensure cosmetics are hypoallergenic is one thing. Teaching a species to be our equals sound completely ethical to me. Let's do it.
posted by lostguy at 10:41 PM on June 17, 2018

Clawsoon, I suspect various CRISPR techniques might help to speed up a seemingly foolish venture like this. If only birds could be cloned? African grey parrots can talk and express themselves pretty well, so I could see a very optimistic and lucky breeding program get a reliably talkative and thoughtful new species of parrot in a not-so-dismal timeframe, especially if it could be done in parallel instead of a serial set of trial and error experiments. Maybe parrots could be trained by robots to eliminate some variables. Perhaps cyborg parrots will be our Singularity. Or maybe I've watched too many Rick and Morty episodes. Didn't Morty's pet dog become a sentient cyborg?
posted by lostguy at 10:59 PM on June 17, 2018

I was also going to suggest that CRISPR techniques, someday, might be helpful, when we better understand what genes are related to intelligence. But CRISPR could deliver consciousness, hypothetically, in the form a shot from a hypodermic needle.

There is a fascinating story about a scientist who, in the process of trying to find a cure for his brother's degenerative disease, discovered a gene that is basically broken in all humans that he hypothesizes is the reason for our comparatively small jaw muscles in relation to other animals. This smaller jaw muscle led to an increase in skull size, which led to larger brains, which led to consciousness. You should be able to find the story on Google.

There is a science fiction show that features a super-intelligent boy, who was made super-intelligent by basically keeping him in a womb-like chamber for years. In the show, it was stated that one theory for why humans are intelligent is their long gestation period. I remember looking this up on google and there was some real scientific theory behind it. I don't remember the show, but you should be able to find it via googling.

There is also the interesting case of the Russian scientist that bred wild foxes into a domesticated breed, which are more social and share many traits we consider necessary for human-like intelligence. This was a big deal, because he did this in a few generations, and, I believe, before that scientists believed it would take tens of generations.

One animal that scientists believe may be as intelligent as humans (other than dolphins) is the cuttlefish. There are some amazing videos online of its ability to blend / disappear into its surrounding environment, an ability that leads some scientists to believe the cuttlefish's brain must be managing incredible complexities. Who knows, but it's really interesting.

Another note on CRISPR. I was listening to a Sam Harris podcast with the woman who discovered / invented CRISPR and he asked her whether or not she was worried that people would use it to do things like design more intelligent babies. She said that would be complicated and society would probably ease into that slowly. He replied that he rather believed people would do it pretty much as soon as they could. She ... immediately caved and said she agreed. I was stunned by that. Also, she described the process by which someone can change their eye color from brown to blue, pretty much a shot from a hypodermic needle, and then a wait of a couple of hours. This podcast was a very good introduction to CRISPR. Vox also has an explainer article or video, not sure which.

Good luck with your experiment ;-)
posted by xammerboy at 1:08 AM on June 18, 2018

This article may be of interest

The Faculty of Language: What Is It, Who Has It, and How Did It Evolve?
posted by moorooka at 2:32 AM on June 18, 2018

You might want to read the novel Flowers for Algernon. It was made into a movie called Charly, for which Cliff Robertson won best actor in 1968. Algernon is a mouse who receives surgery that increases his intelligence. The surgery is then done on Charly, who is mentally challenged, but becomes a genius.

I’m still not convinced that this is worth animal suffering - and it would be suffering. Living in a laboratory by itself is suffering. But any experiments along these lines are going to involve killing the animals and dissecting their brains. Almost nobody really looks into the lives of laboratory animals. Even under the least cruel circumstances, it’s not pretty.
posted by FencingGal at 5:46 AM on June 18, 2018

If it's not become clear, one of the big changes in our thoughts regarding animals is that consciousness is that consciousness is now regarded as more of a continuum. Mammals share a lot of our brain structure, many believe too much to not have a consciousness. While science cannot prove animals have many intelligent characteristics, it's obvious to most anyone they do have these characteristics. For instance, science has found it difficult to prove that animals feel pain. Your average pet owner would be astonished at some of the things science doesn't accept in regards to animals. Of course, their intelligence need not be the same as ours as well.

Some religions regard consciousness itself as a form of suffering. Science, on the other hand, has found a direct correlation between happiness and success in life and I.Q. If you were dissecting animals brains to see if they were becoming more intelligent, as of now no one would know what to look for. Until recently, scientists pretty much believed bigger brains meant more intelligence, but some recent evolutionary finds, for example of the "hobbits" suggest that much smaller humans with much smaller brains were capable of using tools. A recent finding along these lines suggest their brains were structured in complicated ways as well.

Intelligence is pretty difficult to define, hence the Turing test, which pretty much states that if a computer's behavior is indistinguishable from an intelligent human's that's good enough. This is the same standard that science has been unwilling to accept as evidence of intelligence in animals, where there seems to be a stubborn resistance to accepting demonstrated behavior as genuine rather than simply mimicked, taught, etc. That said, I don't see any reason why breeding for intelligence need be cruel at all. Humans do it just by marrying someone they feel is smart.

Flowers for Algernon was originally an award winning short story if you don't feel like reading the book. I cried after I read it. I was 12, but even thinking about this story makes me want to cry. There's a movie (Pictures with Charly?) and even that's really good.

There are a lot of scientists that work with monkeys to try and unravel how the brain works. Some of the experiments are undoubtedly cruel. A lot of undoubtedly good things have come from the experiments as well - cures for some forms of blindness and so on.
posted by xammerboy at 10:04 AM on June 18, 2018

Response by poster: Lots of good food for thought here. I think if computer scientists are going to try to create AI with general artificial intelligence, there should be an analogous group trying to do it "organically" too.

Thanks for the thoughtful discussion, as usual.
posted by mhh5 at 3:31 PM on June 19, 2018