Rosalind Arden, when not sitting next to me at conferences teaching me what questions to ask speakers about animal general intelligence, has been testing dogs on doggy problem-solving, and finds that the general ability evident in mice, chimps and humans is also present in dogs. Odd if it were not the case, given that it is a feature of having a brain which must adapt to the problem of survival.
I accept that dogs co-evolved with humans, and do not resent them their place on the planet, but I prefer most of them at a distance, outside the house and in the countryside where they belong. Once I am sure that they are not going to bite me, lick me or piss on the furniture I can tolerate the quieter and more thoughtful ones, particularly if they are morose but amiable. Excitables dogs, overjoyed that humans have genitalia, I can do without. Cats, with their claws, and their tendency to sit in your lap, I can also do without. Perhaps in my case I was left out of the co-evolution process.
Rosalind Arden and Mark James Adams. A general intelligence factor in dogs. Intelligence Volume 55, March–April 2016, Pages 79–85
The structure of cognitive abilities in dogs is similar to that found in people.
Dogs that solved problems more quickly were also more accurate.
- Dogs’ cognitive abilities can be tested quickly, like those of people.
- Bigger individual differences studies on dog cognition will contribute to cognitive epidemiology.
Hundreds of studies have shown that, in people, cognitive abilities overlap yielding an underlying ‘g’ factor, which explains much of the variance. We assessed individual differences in cognitive abilities in 68 border collies to determine the structure of intelligence in dogs. We administered four configurations of a detour test and repeated trials of two choice tasks (point-following and quantity-discrimination). We used confirmatory factor analysis to test alternative models explaining test performance. The best-fitting model was a hierarchical model with three lower-order factors for the detour time, choice time, and choice score and a higher order factor; these accounted jointly for 68% of the variance in task scores. The higher order factor alone accounted for 17% of the variance. Dogs that quickly completed the detour tasks also tended to score highly on the choice tasks; this could be explained by a general intelligence factor. Learning aboutg in non human species is an essential component of developing a complete theory of g; this is feasible because testing cognitive abilities in other species does not depend on ecologically relevant tests. Discovering the place of g among fitness-bearing traits in other species will constitute a major advance in understanding the evolution of intelligence.
However, though I am not interested in dogs, I can see why they make good experimental subjects. The authors say:
Dogs are not subject to confounding arising from lifestyles that may contribute to causal differences such as smoking, alcohol and drug use. Individual differences in dogs’ cognitive abilities are not causally confounded with variability in socio-economic status. It is more feasible, cheaper and less intrusive to conduct repeated behavioural testing with dogs. Following phenotypic studies, dogs will be useful in genetic studies; genes associated with complex traits are easier to find in dogs than people because of their longer haplotype structure (Lequarré et al., 2011 and Ostrander et al., 2006). A consequence of their haplotype structure is that sample sizes needed for genomic analyses are much smaller in dogs than people. Some behavioural adaptations are breed-specific (pointing, herding); these involve both innate propensities and learning. Some traits are typical across all breeds, such as a tendency to affiliate with humans (see for review Benksy et al., 2013, Miklosi, 2007 and Shipman, 2010).
Here are the doggy tests they used: We examined individual differences on a set of cognitive tasks (four increasingly complex versions of a detour task first designed in 1927 by the German psychologist, Wolfang Kohler (1887–1967)(Frank and Frank, 1982 and Scott and Fuller, 1965), a quantity-discrimination task (Bonanni et al., 2011,Macpherson and Roberts, 2013, Prato-Previde et al., 2008 and Ward and Smuts, 2006) and a point-following task (Elgier et al., 2012, Ittyerah and Gaunet, 2009, Kaminski and Nitzschner, 2013, Lakatos et al., 2012 and Miklosi et al., 2006). These tasks were administered to one breed of dog (border collies) selected from similar rearing and living environments. We administered six tasks (of which four were related) to the dogs and, guided by the human psychometrics literature, tested the fit of four basic models against the data.
We recruited 68 farm-living border collies from Wales. We chose a single breed to avoid confounds arising from differential selection. Scores from a basset hound tested against a whippet would be uninterpretable (Udell, Ewald, Dorey, & Wynne, 2014) This is because dogs have been selected by people for different behaviours, and they are the most polymorphic species on earth, varying greatly in leg length and other traits relevant to task performance. 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 level of enrichment could contribute to cognitive differences). 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 reported moderate inbreeding coefficient of around 2.8% (Hoffman, Hamann, & Distl, 2002) but unknown empirically in our sample. 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.
The animals in our sample differ from companion animals in background and behaviour that may be relevant to the study. They are kennelled outdoors and, although socialised to respond to their owner in a farmyard setting, they are unaccustomed to games, indoor behaviour and food treats.
We first estimated how much within-dog variability there was on task performance. The consistency of performance was low for navigation (R = 0.26, 95% credible interval [CI] = 0.11, 0.42) and repeatability was low for the point-following (R = 0.35, CI = 0.22, 0.50) and moderate for quantity discrimination (R = 0.51, CI = 0.40, 0.63). Consistency on mean navigation completion time was moderate (R n0 = 0.58, CI = 0.35, 0.74). Repeatability of mean completion time on point-following was high (R n0 = 0.77, CI = 0.63, 0.87) and of average completion time on quantity discrimination was also high (R n0 = 0.88, CI = 0.83, 0.92).
The authors then carry out some factorial studies, and found that the hierachical g model was the best fit. The full model explains 68% of the variance, but g itself accounts for only 17% of the variance, which is not all that much, but see the possible explanation further below.
Our results indicate that even within one breed of dog, where the sample was designed to have a relatively homogeneous background, there is variability in test scores. The phenotypic structure of cognitive abilities in dogs is similar to that found in people; a dog that is fast and accurate at one task has a propensity to be fast and accurate at another. It may seem obvious that once a detour task (finding the treat behind a barrier) has been solved in one form, the solution to the other forms will follow naturally, but dogs are not people. Experiments have shown that dogs’ problem-solving skills do not transfer readily from one problem to a different form of the same problem as ours do (Osthaus, Marlow, & Ducat, 2010). The g factor we report is consistent with the prediction made by the many experts in the ‘dog world’ (trainers, veterinarians, members of dog societies, and farmers) who were consulted in the early stages of this study. Those experts said that in their experience some dogs were more likely to catch-on, learn and solve problems more quickly than others. Our results show structural similarities between canine and human intelligence.
Just as everything is falling nicely into place, Arden and Adams admit that border collies which can’t round up sheep at the command of the shepherd do not make the grade as sheep dogs, and are excluded, so that the sample they’ve been working with are the cognitive elite of the breed, able to work out what humans want them to do, the top flight university graduates of the doggy world. This selection will lead to considerable restriction of range, because all the dogs are very bright. If it were possible to include the equivalent of some technical college collies, a few art college collies, and a pack of liberal arts, humanities, social scientist qualitative sociologist collies, it would strengthen the correlations and the g factor considerably. Perhaps this should be incorporated in the next study. The other approach would be to find a harder additional task to test these bright collies more stringently.
Read the whole paper here:
This is an important finding, and will not be surprising to biologists, but certainly dents the argument that general ability is a confection which depends on an arbitrary selection of narrow tests.
So, Deary and Der’s “cognitive epidemiology” is now joined by Arden and Adams’ “dognitive epidemiology” and g rides over the whole lot, supreme.