Tuesday 10 December 2013

ISIR Press Release: Like humans, dogs have general intelligence

 

from the International Society for Intelligence Research Conference in Melbourne Australia. NO EMBARGO.

General intelligence in dogs

Arguments have raged as to whether human intelligence contains a large general component, or whether that just arises from the way that IQ tests are constructed. Now Dr Rosalind Arden, Dr Mark Adams and Prof Robert Plomin have tested the intelligence of border collies on four different detour route-finding tasks and two different tests of pointing.

They say: A higher order factor accounted for 40% of the reliable variance in performance. Dogs that quickly completed the detour tasks also tended to score higher on the choice tasks and take less time to make a choice, and this could be explained by a general intelligence factor. The weaker inter-correlations among test performance implies that dog intelligence may have evolved in response to selection for specific behaviors and abilities.

So, if you consider that general intelligence account for 50% of human ability, and 40% of border collie ability they have much in common in terms of mental organisation, but dogs are more likely to have some specific abilities because they were bred up to do specific tasks by humans.

Contact: Dr Rosalind Arden   Rosalind.arden@kcl.ac.uk

If she is on a plane at the moment, I can provide a preliminary summary.

11 comments:

  1. Anyone who doubted that Border Collies are intelligent doesn't know them. Anyone who doubted that some are more intelligent than others, ditto. To confirm that their g is like our g, you just need to study its heritability, which should be easier with collies.

    And now the big questions: what was the sample size, and how was the sample drawn?

    ReplyDelete
    Replies
    1. We recruited 84 male and female farm-living border collies from Wales (N=68 dogs completed all tests). We chose a single breed to avoid confounds arising from differential selection (scores from a basset hound tested against a whippet would be uninterpretable). We selected farm border collies for several reasons. First, we wanted the dogs’ backgrounds to be similar (in contrast with pet, or companion animals because variation in levels of enrichment could contribute to cognitive differences). Some farmers own several dogs; this reduced further the variation in home environments across dogs in our sample. Although border collies have been subject to artificial selection its focus has been on behaviour more than appearance; border collies remain morphologically variable with a moderate inbreeding coefficient of around 2.8% (ref). Our sample comprised 68 dogs, (males 34, females 34) ranging in age from 1 to 12 years. We chose Wales as our recruitment centre because it is rural and enriched for border collies, having many hill farms where dogs work stock.

      Delete
  2. Lots of interesting stuff at the ISIR conference. The program, with abstracts, is here.

    ReplyDelete
    Replies
    1. Can't wait for "Petaflop Computer Clusters Crunch Genomes of Giant Minds:
      Collection, Sequencing, Analysis of the World’s Largest Extreme-IQ Genomic Cohort":

      "We describe the model of the genetics of intelligence, variance, and mutation used in the study design, the reasoning underlying this model, the implications on what types of variants we look for, and the prospects for the collection of further, equivalent cohorts in East Asia and Scandinavia."

      Brainiac Hsu has a video about the genetic architecture of intelligence: http://www.youtube.com/watch?v=FgCSkGeBUNg but beware, one result did not replicate (the one about differences in genetic distances between the high IQ and normal group)

      Delete
    2. Another good one (additivity is king. Epigeneticists wept.):

      "The evidence for the pervasiveness of small-effect architectures is now strong enough that we propose a "Fourth Law of Behavior Genetics": genetic variation in a typical behavioral trait is attributable to thousands of variants scattered across the human genome, each of which has a small effect. Some commentators have expressed disappointment at these small effects, which are denigrated as academic curiosities with little enduring scientific value. At first blush these claims, which cite the writings of authorities such as Paul Meehl, seem to have some merit and thus appear to throw the entire enterprise of gene hunting into disrepute. In this lecture I examine the epistemological implications of the Fourth Law and conclude that gene-mapping studies are in fact scientifically well justified."

      Delete
    3. So exciting! One more:

      "In 1988 (conducted 1984), Snyderman and Rothman published a survey of
      expert opinions on intelligence and its treatment by the media. Since their survey, new findings have appeared in the literature. To examine current opinions of experts, we created a new questionnaire (“Expert Questionnaire on Cognitive Ability“), partly based on the questions of Snyderman and Rothman, reflecting the current state of research. We added questions on contemporary topics such as the FLynn effect and international intelligence comparisons. Other questions addressed the definitions of terms (e.g., intelligence and cognitive ability); the validity of tests (e.g., WAIS/WISC, SAT, PISA); the estimation of genetic and environmental influences; the treatment of intelligence research by the media; the importance of genetic testing (e.g., DNA screening); and the development of intelligence in global regions and populations."

      Will these be released as papers or what?

      Delete
    4. Endre, you are stealing my thunder, and racing ahead of me! (for which many thanks). I am not sure that we will get much out of the BGI group, because they are still crunching the data, apparently. Rindermann should be making some of the results of his survey available to me soon, but since the whole gang will be in a lecture hall at 8 am tomorrow I doubt we will get much out of them for a few days.
      I am sorry to miss this conference, but Melbourne is a long way away.

      Delete
  3. I rembered the name Arden because I thought a quip of hers a few years ago was very clever "...This does not mean that men who prefer Play-Doh to Plato always have poor sperm.."

    http://news.bbc.co.uk/2/hi/health/7767877.stm

    ReplyDelete
    Replies
    1. Wouldn't "Rosalind Arden" be a great name for a Shakespearean heroine?

      Delete
    2. Kind of like in Plomin's new twin study paper, the contact author is Nicholas Shakeshaft.

      Delete
  4. Rosalind Arden is a heroine as far as I am concerned! She sat next to me throughout most of the last ISIR conference, and fed me comments to include in my questions to speakers, thus creating the impression that I knew all the methodological problems involved in animal learning experiments.If I had been called Shakeshaft, however, I would have needed no help from anyone.
    By the way, I am just reading the results of the survey on intelligence. Please read that post when it comes out. You will be interested.

    ReplyDelete