The concept of 'intelligence' within the biological sciences.

[Nereid]

Recently, the Biology section of PF got a new thread, Cranial Size and Intelligence. One exchange in that thread concerned the 'intelligence' of species other than Homo sap.

Aside from comparison within humans. Do you believe there is a cranial size to intelligence correlation when comparing species to species?

I have no idea how one could objectively determine the 'intelligence' of another species! Perhaps you could help?

Really? So you're stating one will have a difficult time determining which is more intelligence. Dolphins or flies. Wolves or salmon.

For example, how does one measure the intelligence of Fagus grandifolia, or Deinococcus radiodurans?

Perhaps a little common sense.

Clearly, not wearing a biologist's hat, each of us may be able to give a description of what we think the intelligence of another species might be, and these descriptions might have some common elements.

But how do biologists use the term? For example, does it have a set of precise, technical meanings? Is there a set of standard techniques regularly used to measure the intelligence of species? Are there objective scales for intelligence (within the biological sciences)?

Ernst Mayr, the famous evolutionary biologist, certainly used the term in his http://www.planetary.org/html/UPDATES/seti/Contact/debate/Mayr.htm . However, there he was addressing a non-technical audience, and likely used the word in its everyday meaning, not necessarily in any precise, biological sense.

Can any biologist help here?

 

[hitssquad]

The 10 trillion dollar comprehensive investigation into an interspecies g factor

concerned the 'intelligence' of species other than Homo sap... each of us may be able to give a description of what we think the intelligence of another species might be ... But how do biologists use the term?

Arthur Jensen taught biology and genetics courses at UC Berkeley and claims (Miele & Jensen, 2002) he is an expert in biology. Here is his description of intelligence in others species:

• The main indices of intelligence in animals are the speed of learning and the complexity of what can be learned, the integration of sensory information to achieve a goal, flexibility of behavior in the face of obstacles, insightful rather than trial-and-error problem-solving behavior, transfer of learning from one problem situation to somewhat different situations, and capacity to acquire abstract or relational concepts.

(Arthur Jensen. Bias in Mental Testing. From the section "Animal Intelligence" (pp 175-182).)

Is there a set of standard techniques regularly used to measure the intelligence of species?

To name a few, there is the Morris water maze, the radial maze, and the cross-maze. Abstracts and references for over 40,000 investigations into rate of rat learning are currently indexed on PubMed (and those hits only include the titles or abstract that specifically mention learning; some investigations of rat learning rates may not mention learning by name). The techniques used to quantify rate-of-learning are usually referred to in those abstracts.

Are there objective scales for intelligence (within the biological sciences)?

I do not know of any that are cross-species. I think an objective interspecies scale of general mental ability would have to consist of an interspecies g factor extracted from tests given across species (e.g., the g factor extracted from tests of mental ability administered in the United States is not the same g factor as one extracted from tests of mental ability administered in sub-Saharan Africa; but it is virtually the same, as objectively established by techniques described by Jensen in The g Factor, and in order for a g factor that applied to both to actually be the same, a single factor general to both cultures would have to be established and would necessarily account for the less of the predictive variance within each culture than general factors of mental ability specific to each culture would account for (e.g., if 40% of variance in social outcome within Culture A is accounted for by its within-culture g, 39% might be accounted for by the between-culture g shared by Culture A and Culture B)). Some tests would require translating between species with proxy tests in order to bridge the gaps between necessarily different testing procedures between different species (not impossible, at least between human cultures, according to Jensen (1998)) and it would necessarily be limited in generality only to the animal species tested (i.e, the more species-comprehensive any given study would be, the more general the interspecies g factor extracted from the tests administered in that study would be; the established g factor might then be usable as a qualifier for intraspecies mental abilities tests in terms of their respective loading to objectively quantifiable degrees on this interspecies g factor). I would guess that the reason we do not have one is that the need for objectively and quantitatively measuring differences in general interspecies mental ability has never arisen. When tests are performed involving administration of potential IQ-raising or neuroprotective chemicals to animals, it is only the relative differences within the small group of animals tested (and their controls) that is relevant to the study at hand. Animals in studies such as these are always of the same species, as far as I have been able to see from the 10,000 or so biological abstracts I have skimmed, hence biologists have to need for an aobjectively quantified interspecies g factor of mental ability.

On the other hand, if a biologist decided to investigate the efficacy of one or more mental-ability-raising techniques to raise the general mental ability level of a group of animals of one species over that of another species, then he would seem to certainly need a measure of mental ability general enough to apply equally to both species.

 

[Monique]

Intelligence is such an unclear term to define, especially across species. There is a term called encephalization quotient: a number reflecting the increase in brain size over and beyond that explainable by an increase in body size.

 

[hitssquad]

Mental ability hard or easy to generally define across species

Intelligence is such an unclear term to define, especially across species.

Mental ability seems to be clearly definable across species:

• Item Performance (IP). This term, henceforth abbreviated IP, refers to any distinct voluntary behavioral act. It can be any such overt action provided it is also observable or recordable in some way. Saying (or writing) "four" (or any other response) to the question "what is two plus two?", if the response can be agreed upon by observers, is an IP. Doing a triple axel, writing one's name, hitting middle C on the piano, performing the Tchaikovsky violin concerto, solving (or failing) an attempted math problem, jumping over a two-foot hurdle, cutting an apple in half, hitting a baseball into left field, and parking parallel-these are all IPs. An IP may be a discretely classifiable act (e.g., either hitting or missing a target) or some action that can be graded on a continuum (e.g., speed of response to a signal, time taken to complete a task, distance run in a given time). The universe of possible IPs is obviously unlimited. The definition of IP also includes, of course, a voluntary response to an item in any kind of test or to any laboratory procedure that measures, for example, reaction time, sensory threshold, speed of rote learning, or memory span.
  
  Excluded from the category of IPs are unconscious, involuntary, or accidental acts, such as tripping on a stair, eye blinks, facial tics, unconditioned and conditioned reflexes, reactions of the autonomic nervous system, somnambulistic actions, drug reactions per se, fainting, and the like. Organismic events that are not strictly behavioral acts are also excluded, such as changes in brain waves, glandular secretions, pulse rate, blood pressure, skin conductance, and pupillary dilation, although these phenomena may be correlated with certain IPs. Most importantly, it should be noted that an IP is not an inference, or an abstraction, or an interpretation. It is a single objective raw datum – some overt act, directly observable by other persons or immediately recordable by an apparatus.
  
  Ability. Going from an IP to an ability is going from a direct observation to an abstraction or inference, although of the lowest order. The universe of abilities is open-ended but bounded by certain qualifications.
  
  An ability is an IP that meets the following three criteria:
  
  
  • (1) it has some specified degree of temporal stability (consistency or repeatability); (2) it can be reliably classified, measured, ranked, rated, graded, or scored in terms of meeting some objective standard of proficiency; and (3) it has some specified degree of generality.

(Arthur Jensen. The g Factor. pp50-51.)


To see how it relates within and between groups, we use constructs called factors:

• Factor. The word "factor" has a number of dictionary definitions, but the term as used here has a very restricted, specialized meaning. A factor is a hypothetical variable that "underlies" an observed or measured variable. Thus a factor is also referred to as a latent variable. It is best thought of initially in terms of the mathematical operations by which we identify and measure it.
  
  Although a factor is identifiable and quantifiable, it is not directly observable. It is not a tangible "thing" or an observable event. So we have to be especially careful in talking about factors, lest someone think we believe that we are talking about "things" rather than hypothetical and mathematical constructs. But one can say the very same thing about the many constructs used in the physical sciences (gravitation, magnetism, heat, valence, and potential energy, to name a few). They are all constructs. This does not imply, however, that scientists cannot inquire about the relationship of a clearly defined construct to other phenomena or try to fathom its causal nature. Nor is a construct "unreal" or "chimerical" or less important than some directly observable action or tangible object. Certainly the force of gravity (a hypothetical construct) has more widespread importance than the particular chair I am sitting in at the moment, and is every bit as real.

(Arthur Jensen. The g Factor. pp55-56.)

There is a term called encephalization quotient: a number reflecting the increase in brain size over and beyond that explainable by an increase in body size.

Mental ability has nothing directly to do with biological artifacts (such as brain size):

• A lot of pointless arguments can be avoided by consistently maintaining a clear distinction between the purely mathematical definition, identification, or measurement of factors, on the one hand, and theories about their causal nature, on the other. (In this chapter, I am not saying anything at all about the causal nature of factors.)

(Arthur Jensen. The g Factor. p56.)


If mental ability was inextricably linked to biology, artificial intelligence would be impossible:

• Q. Are computers the right kind of machine to be made intelligent?
  
  A. Computers can be programmed to simulate any kind of machine.
  
  Many researchers invented non-computer machines, hoping that they would be intelligent in different ways than the computer programs could be. However, they usually simulate their invented machines on a computer and come to doubt that the new machine is worth building. Because many billions of dollars that have been spent in making computers faster and faster, another kind of machine would have to be very fast to perform better than a program on a computer simulating the machine.

 

[Moonbear]

To name a few, there is the Morris water maze, the radial maze, and the cross-maze. Abstracts and references for over 40,000 investigations into rate of rat learning are currently indexed on PubMed (and those hits only include the titles or abstract that specifically mention learning; some investigations of rat learning rates may not mention learning by name). The techniques used to quantify rate-of-learning are usually referred to in those abstracts.

Biologists are very cautious about applying the term "intelligence" to what is determined by any of these tests. We do use them to indicate learning and/or memory, but whether performance on any of those tests can actually tell us anything about the intelligence of an animal, I don't really think so. They are usually used to compare responses to a particular treatment with a control. For example, one might use the Morris water maze to compare the ability of a group of animals treated with a dopamine antagonist to controls to determine if the antagonist affects learning or memory. There is no way to distinguish which is affected, whether the animals don't learn the task, or if they aren't able to recall what they've learned. If the test was given alone, it also wouldn't be able to distinguish if a slower time to find the platform was due to an overall motor dysfunction (i.e., they remember where the platform is, but just can't swim as fast to get there). Usually a battery of tests is used for such studies to help narrow down where the deficits are occurring. For example, the ability to walk across a rotating dowel, or putting ink on their feet and letting them walk across a piece of paper to measure their stride to ensure motor coordination is intact. Poor performance can also relate to problems in sensory function. These tasks usually involve visual (white vs black walls, or different patterns, or open sides vs closed sides) or tactile cues (different types of flooring, i.e., wire mesh vs bars) to help the animals differentiate different arms of a maze or location in the water tank. If for some reason they can't make that distinction, there is little chance they can tell one arm of the maze from another. And these can be affected by other behavioral states, such as fear and anxiety. For example, an animal trained that there is a treat at the end of one arm of a maze may not respond quickly to find it if a treatment they receive makes them more fearful or anxious so that they don't venture out into any arm of the maze and just stay put in the center platform. Or, in the water-maze task, if the platform is placed in the center somewhere, if the animal is very anxious, they may stay all along the edge of the tank and never venture toward the center (these tests also have time-limits so the animal doesn't drown).

Hitssquad is right that it would be very difficult to extrapolate results between species. To some extent, we do. We use the performance of a rat given a drug to tell us something about how we expect humans to react to the same drug. However, this always must be done with caution. Not all animals can be compared the same way. Each of the tests used for a species relies on species-typical behaviors. In terms of learning and memory tasks, we have to rely on the abilities of animals, and cannot directly test what they know.