Home > Centre for Philosophy of Natural and Social Science > Who's Who > Staff > Rosalind Arden > The Serious Science behind the Dog Intelligence Research Programme: Dogs and Dementia

The Serious Science behind the Dog Intelligence Research Programme: Dogs and Dementia

Dementia is an avalanche up the hill that threatens to bury the next generation; this is a consequence of the global shift in the ratio of young to old people. We are living longer, and older age is the largest cause of dementia. The personal and economic costs of this disease are immense. At present we cannot identify those at greater risk, have no treatments, and limited knowledge about its causes. We need novel approaches. I propose an ambitious programme of research that will deliver a full natural animal model of dementia.

In humans the two predictors of later dementia are the ε allele of APoE gene and, independently, lower pre-morbid cognitive ability. It is unknown whether this latter relation is causal; further it is harder to test in people owing to confounding arising from variation in education, income, and lifestyle.

Dogs provide an ideal model to test this hypothesis; I will test whether early poor cognition is causally associated with dementia, for partly genetic reasons. Dogs are ideal for this research and they are free from key epidemiological confounders such as drugs, alcohol and smoking. If early cognition increases risk of dementia in dogs, this project will deliver new endpoints relevant to AD (cognitive scores) for pre-clinical studies of interventions including drug therapies.”

Animal models are crucial in understanding the pathophysiology of complex diseases. The go-to model for dementia is the rodent. Rodents do not acquire dementia; dogs do. Dogs naturally reproduce key aspects of human behavioural and neuropathological ageing as well as dementia processes. In dogs, Aβ, tau and tangles are deposited. The canine form of Aβ is identical to ours (unlike the rodent form) as are dogs' APP processing mechanisms. Scores from published dementia questionnaires administered to owners of old dogs correlate (significantly) and positively with dogs' brain pathology at necropsy. Dogs share greater genetic homology with us than do rodents. Dogs' drug sensitivity and tolerance are similar to ours. Dogs' behavioural changes following cognitive decline are also like ours. They include alterations to, the sleep-wake cycle, appetite and interest in social interaction. Researchers have noted this relevant similarity and published findings concerning canine models of ageing and dementia.

The heavy brake on existing work on canine models of Alzheimer's is that it can take up to 40 days training per dog (to a cognitive criterion), prior to assessment. In addition, it is costly to buy, raise and house dogs in the laboratory.  Convenience samples of pets consist of several breeds; this adds noise to the experimental design.

My previous work has shown that I can test a dog's cognitive performance in one hour.  I established an association with the Sheep Dog Society (ISDS). This unique resource enabled me to acquire, for this project, the ages and postcodes of 34,000 dogs in the UK (all the same breed) and map them on the UK. This obviates the need to buy, or house, any dog. By bringing my expertise in individual differences in intelligence into my work with dogs, I can use genomics to assess the association of early canine cognition and dementia within a 5-year study. This will amplify greatly, the usefulness of the canine model of ageing and dementia. The most invasive procedure in this project will be a cheek swab.


Specific research objectives:

  1. Refine our existing canine cognitive test battery.
  2. Administer the test battery to 900 young dogs (<4 years).
  3. Administer a published dementia questionnaire to owners of 900 old dogs (>8 years)
  4. Administer the test battery to a subset of old dogs for validation and power enhancement.
  5. Genotype all 1800 dogs.
  6. Test the association between cognition in young dogs and dementia scores in older dogs using whole genome SNP data in mixed linear models to estimate the genetic correlation between these traits. The sample N arose from our power analyses.


Hypotheses: Cognitive ability, and dementia scores in dogs will be reliably measured, and heritable. There will be a significant negative genetic correlation between these two traits.

The study will create a robust basis for testing whether cognitive performance measured early in life is associated with successful ageing. It will lay the ground for a later longitudinal study, founding a canine brain biobank in collaboration with Brains for Dementia Research, and rich lifetime phenotyping.