In an earlier blog post, I mentioned the animal welfare science and bioethics Journal Club I am involved in. We meet fortnightly to review articles of relevance to animal welfare, behaviour, and bioethics. The intention is to broaden and disseminate knowledge within the research group and question and improve on our existing ideas and methods.
This week, it was again my turn to select and present an article – and of course I chose one involving cats! (it’s no secret that I am a crazy cat lady)
But I also chose this article because I am interested in behavioural profiling (of humans and animals) and wonder if this will be a requirement of animal-based research in the future. Will we have to declare the temperament of our test subjects in the same way as we record their sex and ages?
The article I chose to review was:
Litchfield, C. A., Quinton, G., Tindle, H., Chiera, B., Kikillus, K. H., & Roetman, P. (2017). The ‘Feline Five’: An exploration of personality in pet cats (Felis catus). Plos One, 12(8), e0183455. doi:10.1371/journal.pone.0183455
It is published in an open-access online journal here.
A few words on personality profiling in humans
One of the most widely used personality profiling devices in humans is the ‘Big Five’. It is based on ‘common language’ descriptors of personality. Each ‘factor’ consists of related traits and characteristics that fit within them. An individual’s personality is determined by where they sit on the continuum for each of the five factors.
The five factors are commonly known by the acronym ‘OCEAN’:
- Openness to experience
Who wrote the article and why?
Human psychologists and ecologists in South Australia and New Zealand (Victoria University of Wellington) were involved in the study. It is part of a larger project (Cat Tracker Project) aimed at tracking pet cats in South Australia and New Zealand.
It is interesting that there appears to be no input from animal scientists – behaviour or welfare. A quick perusal of the reference list indicates that a large volume of feline behaviour and temperament literature appears to be overlooked – or at least not referenced. And, as the authors wrote about animal ‘personality’/expressivity, or behaviour profiling, it also seems absurd that Francoise Wemelsfelder’s Qualitative Behaviour Assessment (QBA) work was not referenced.
The Journal Club group also discussed the idea of ‘personality’ in animals. Many of us thought ‘person’-ality was the wrong terminology to use when discussing animal traits. The preferred term is ‘temperament’. It was our view that the human-centric approach taken for this study has resulted in this wording. I only wonder at why the article was not instead named ‘An exploration of ‘purr-sonality’ in pet cats…’ – at least this has an adorable human-centric approach.
The authors set-out to “analyse personality in a large sample of pet cats – using personality inventory completed by the owner”. Their research question was “how many reliable and interpretable factors depict personality in pet cats and what traits do they represent?”. They based these questions on previous research which have used: small sample sizes, wild/captive cats or those with behavioural problems – rather than ‘normal’ cats, researchers evaluating cat behaviour – rather than animal carers, and insufficient inter-rater reliability assessments.
An online survey in South Australia and New Zealand collected information about 2,802 cats. It used 52 items based on a previous survey of Scottish Wildcats. Cat owners were asked to rate their cat on a Likert scale, according to how far they felt each trait was applicable. The authors used a type of exploratory factor analysis called ‘principle axis factoring’ (a technique similar to Wemelsfelder’s ‘principal component analysis’) to understand the relationship among these reported traits. This enabled them to group traits under common headings/factors.
What did they find? (results)
They found FIVE ‘factors’ (common headings for a group of traits) that explained ~47% of the variance in personalities:
Interestingly, the authors chose to deal with items that ‘cross-loaded’, or loaded onto multiple factors, by assigning the item ONLY to the highest loading factor.
Why cat personality?
The authors drew our attention to some benefits of this type of research, namely: grouping compatible cats in multi-cat households and providing more hiding places for neurotic cats. A low/high score in an area could also improve the welfare of that cat by encouraging owners to seek advice.
The main issues we had with this article
- The authors appear to have not considered animal (welfare or behaviour) scientists or a range of feline behaviour references in their article;
- Reference to Francoise Wemelsfelder is sorely lacking – using 52 items from a Scottish Wildcat study also seemed unusual when the intention was to steer away from references to ‘wild/captive’ cats;
- No consideration was given to the human-animal bond and how attached an owner was to their cat and/or how well they knew their cat – could this have contributed to a reporting bias for some owners?;
- The labelling of 4 of the 5 personality factors were reasonably well explained, but the sudden inclusion of the fifth (‘extraversion’) appears contrived. We had to wonder if they were trying to force it to fit ‘Feline Five’ – it was too similar to the human list (see above);
- Assigning items that cross-loaded ONLY to the highest loading factor ignores the complexities inherent in ‘personality’/temperament profiling.
Overall, this study is a great example of the importance of research collaboration. Think outside the (litter) box.