Join 3,433 readers in helping fund MetaFilter (Hide)


How do you measure intelligence across species?
January 25, 2006 8:55 AM   Subscribe

Assume you had a list of different animal species, and you wanted to order it from most intelligent species to least. (1) What measure of intelligence would you use to compare different species--average number of neurons, for example, or ability to solve certain kinds of tests? Or something else entirely? (2) Can you give some examples of how some common farm, food, or pet animals might rank on the list?

And a special bonus question:

(3) Would your ranking end up with clusters of similarly-intelligent species, with each cluster separated by large gaps from the next one? Or would it be more a steady progression from least intelligent to most?

I recognize that there is not yet a scientific consensus on the meaning of "intelligence" within a single species, let alone among different ones, but I'm interested in well-informed best guesses, and even plausible speculation. Also, of course, I'm interested in hearing your answer to any of the above questions, even if you can't answer all of them.
posted by yankeefog to Science & Nature (17 answers total) 2 users marked this as a favorite
 
start here
posted by edgeways at 9:03 AM on January 25, 2006


One problem: intelligence is not a single-dimension concept. And we really can't measure raw intelligence, we can only measure demonstrable abilities. There are many, perhaps unlimited, kinds of intelligences, and your ranking will depend completely on which kinds you choose. So, for instance, "scent analysis and identification" ability could be considered an important element of intelligence for many species. MEasures of such an ability would put dogs way up there, and people at a not very impressive level.
posted by Miko at 9:28 AM on January 25, 2006


One convenient criteria is the ability of an animal, or a class of animals, to be trained to respond to an increasingly complex set of stimuli (in series or parallel) with a repertoire of learned behaviours.

Octopuses score impressively well using this benchmark.

Most animals with a central nervous system and some bilateral symmetry will display some measure of "learning". By comparison, box jellies, equipped with very sophisticated vision systems and highly complex peripheral nerve nets seem to be untrainable. And I don't think it's just stubborness.
posted by meehawl at 9:44 AM on January 25, 2006


Mefi: Stubborn Jellyfish!
posted by Wilder at 9:52 AM on January 25, 2006


Whatever you do, you apparently should not rely on the ratio of brain weight to body weight.
posted by alms at 10:08 AM on January 25, 2006


Yeah, you're going to have to define what you mean as intelligence, first. That's going to give you the criteria you can use in your ordering.
posted by bshort at 11:06 AM on January 25, 2006


I have to completely disagree with what was contained in that link to the brain weight to body weight ratio. There is no reason to think that mice, elephants, and whales are not smarter than humans. Not having opposable thumbs or vocal chords like ours does not mean they are not intelligent.

On the other hand, I would like to see the data in relation to brain weight and spine length.
posted by pwb503 at 11:33 AM on January 25, 2006


There is no reason to think that mice, elephants, and whales are not smarter than humans. Not having opposable thumbs or vocal chords like ours does not mean they are not intelligent.

No one's saying they're not intelligent. Just no where near as intelligent as we are.
posted by electroboy at 12:12 PM on January 25, 2006


No one's saying they're not intelligent. Just no where near as intelligent as we are.

There is no reason to think this either. Without a level playing field, there is no way to judge if humans are more or less intelligent than these other animals.
posted by pwb503 at 1:01 PM on January 25, 2006


alms writes "Whatever you do, you apparently should not rely on the ratio of brain weight to body weight."


But see also.
posted by orthogonality at 2:06 PM on January 25, 2006


A couple things I think they consider in animal studies are the ability to use tools and the ability to use symbols. One study, I remember, had primates in a situation where they could only obtain (whatever it was--say a banana) if they stood on a chair and used a pole to reach it.

Use of symbols, such as language and numbers, is supposedly what separates humans from other species. I think the main challenge to this premise is that some primates have been taught to use sign language. Some argue, though, that they are only repeating a complex learned behavior in a mechanical way, rather than "using language" as we might define it.

In terms of research, that's about all I remember from whatever long-ago grad school class made me read something about it. Here are some other aspects of intelligence to consider (I'm shooting from the hip here):

The ability to learn:
--in classical conditioning (think Skinner, bell, salivate), how many times must the primary reinforcer (food) be paired with the secondary reinforcer (bell) before the secondary reinforcer (bell) alone elicits the target behavior (salivating)?
--in operant conditioning (think trained seals), over how many "shaping" episodes must the reinforcer (a tasty fish) be provided to go from a state of "never demonstrates balancing-ball-on-nose behavior" to "consistently balances-on-nose when ball is presented?"

The ability to solve novel problems (a lot of IQ testing in humans involves this).

The ability to bring about a desired outcome. For example, one of our cats has learned that standing on my husband's chest and screaming will reliably result in getting fed, even at 5 a.m.

I know this doesn't cover a lot of ground, but perhaps it's a start. Interesting topic!
posted by lisaj32 at 4:28 PM on January 25, 2006


Without a level playing field, there is no way to judge if humans are more or less intelligent than these other animals.

Yeah, no kidding. When it comes to trampling, elephants beat humans every time. Likewise: when it comes to hiding, mouses are tops.

Intelligence isn't a very meaningful concept to start with, and I think questions like this are nearly meaningless if you're not going to specify what you mean by intelligence.
posted by ikkyu2 at 9:28 PM on January 25, 2006


in·tel·li·gence Pronunciation (n-tl-jns) n.
1.a. The capacity to acquire and apply knowledge.
b. The faculty of thought and reason.
c. Superior powers of mind.
posted by lisaj32 at 12:28 AM on January 26, 2006


Sorry. My question was murky and imprecise because my understanding of this subject is murky and imprecise. That's why I'm asking--I'm trying to get a sharper grasp on the topic.

Based on my observations of various species, it seems to me intuitively obvious that my brain is more powerful than a fish's brain, and it seems pretty likely to me that a dog's brain is somewhere in between. I'm looking for help in shaping this unscientific intuition into some sort of more objective measure, which could then be applied across multiple species.

So maybe a clearer way of asking my question might be:

A. Tell me a definition of "intelligence" that you consider to be meaningful, or as meaningful as it can be, given the fuzziness of the concept.
B. Tell me of some way of measuring relative degrees of "intelligence," as well as it can be measured given the differences between species.
C. Tell me how different species rank according to this definition.


I'm not expecting some sort of concrete and uniformly agreed-upon answer. Like I said, I'm happy with best guesses and plausible speculation. Edgeways, Meehawl, alms, and lisaj32 are exactly the kind of answers I'm looking for, if that helps clarify what I had in mind.
posted by yankeefog at 5:53 AM on January 26, 2006


in·tel·li·gence Pronunciation (n-tl-jns) n.
1.a. The capacity to acquire and apply knowledge.
b. The faculty of thought and reason.
c. Superior powers of mind.


That looks like a dictionary's attempt at defining human intelligence. [And note, it's a very simple, superficial definition that reasons from its own premise]. suppose if you want to rank animals according to how closely they approximate human intelligence, you have your question.
posted by Miko at 6:26 AM on January 26, 2006


It sounds like you could frame one part of your question as "How can we operationalize the construct 'intelligence' so that it can be measured across species?"

When we're dealing with constructs--whether "happiness," "extroversion," "the soul," or whatever, things are always going to be a bit murky because they are not tangible objects or uni-dimensional processes that can be easily observed or measured. We want "intelligence" to be less murky (and perhaps we are somewhat miffed that it is not), but there it is.

Because constructs are murky, when we perform research that uses them, they have to be operationalized. That is, we have to define them (for the purpose of our study) in ways that can be observed and measured. Also, we often have to come at them from a number of different directions in order to keep our operational definitions valid, or consistent with how the construct is used in the real world. That is, we measure lots of different things, each of which we think is an aspect of the construct. In studying intelligence, we might use a measure of problem-solving, a measure of processing speed, a measure of memory, and so forth. We then use multivariate (multi-variate = many variables) statistics to make sense of our data. For example, if seven of our variables "clump together" (are correlated) and the eighth does not, we might decide that the eighth variable is not part of this construct--it is not a good operational definition of some aspect of it. So we chuck it.

Ok. To study intelligence across animal species, we will have to measure a number of different observable behaviors that we hypothesize are part of the construct "intelligence." In doing so, we have to take into account the fact that animals with opposable thumbs can do things that animals with no hands at all cannot do. So we want to include behaviors that most species can perform, and when we use a behavior that some species cannot physiologically perform, we will want to control for that when we look at the results. That is, we will "factor it out" or "hold it constant" when performing our statistical analyses. (I think we do that by using the average score from our group of animals that can perform it when we do our calculations, but I'm not sure. The point is, this is a problem that can be managed using statistical means. We don't have to chuck every behavior that some species can perform and others cannot.)

I think I'd better stop here. I hope what I've accomplished with this diatribe is to give you a sense of the complexity of your question, while at the same time suggesting that the question need not be dismissed due to its complexity. The concept "intelligence" is not meaningless: we use it every day, and most of the time we know what we mean.

I think you've asked a fascinating question. :)
posted by lisaj32 at 8:32 AM on January 26, 2006 [1 favorite]


Thanks for the great and thoughtful answers, everybody! You've given me a lot to chew on.
posted by yankeefog at 9:48 AM on January 28, 2006


« Older How much brutal honesty betwee...   |  Good U. S. Supreme Court histo... Newer »
This thread is closed to new comments.