In 1976, psychologists John and Sandra Condry of Cornell University had 204 human adults view videotaped footage of an infant boy named David and infant girl named Dana, and asked them to describe the infants' facial expressions and dispositions. They described their findings in an article in the journal Child Development. In the video, infants were shown responding to various stimuli, which were not visible to the viewer. For example, they'd be shown a teddy bear, so that their reaction could be recorded. They were also videotaped responding to a loud buzzer and to a jack-in-the-box. Participants described David's response to the jack-in-the-box, for example, as "anger," while they described Dana's response to the same toy as "fear." Participants rated David's emotional responses to all three stimuli as more "intense" than Dana's.

Here's the catch: David and Dana were the same infant. Each of the experiment participants were shown the same video of the same infant. Half of them were told the infant was a nine-month-old boy named David, and half were told the infant was a nine-month-old girl named Dana. That they described the "two" infants in such different ways was evidence that the participants' perceptions were at least based in part upon pre-existing biases and preconceptions about the different ways in which boys and girls experience the world.

Now, a group of researchers from Tokyo and Berlin have published a new finding about the relationship between personality and genetics in captive elephants. They collected genetic information from the blood, feces, tissues, cheek swabs, or hair of 196 Asian (Elephas maximus) and African elephants (Loxodonta africana) in Japanese, American, and Canadian zoos, and sanctuaries in Thailand. Personality information was collected for a seventy-five of those elephants by distributing to questionnaires to their keepers. Each elephant was assessed by more than one keeper.

An improved understanding of elephant personality would be not only extremely interesting from a basic science perspective, but also extremely useful for more effectively maintaining captive elephant populations in zoos and sanctuaries. The better that zookeepers and curators understand the psychology of the animals in their collections, the better the quality of care can be, which directly impacts animal welfare.

Every personality psychologist probably has their own nuanced understanding of the word, but a reasonably broad definition of personality might be something like "an observable pattern of thoughts or behaviors, ultimately derived from a set of genes, that an individual expresses consistently over time, in a variety of situations." In other words, personality is reliable: it holds constant across environments, such as home, school, or work. And stable: it doesn't change significantly as an individual ages.

That genetics underlies the stable behavioral patterns of personality is not a particularly controversial idea. Nor is the idea that non-human animals and humans alike have personalities. It is also not particularly controversial to collect personality information from third parties rather than from the individuals who are being studied. For example, in studies of infant or child personality, researchers often distribute surveys to parents or teachers. Indeed, the parent-child or teacher-student relationship is roughly analogous to the keeper-animal relationship at a zoo.

The researchers claim to have detected a correlation between a variant of a gene called ASH1 and neuroticism, for Asian elephants. They also found genetic polymorphisms - variations in composition - in two other genes that they looked at: AR and NUFIP2, but they did not detect correlations for these genes with personality. To be clear, this was not a GWAS, or genome-wide association study, a method plagued by false positives (added to clarify: when sample sizes are small. This sort of candidate gene study has its own share of problems as well. This post focuses therefore on issues surrounding personality assessment, leaving criticism of the genetic methods to those more qualified). They looked only at seven regions on six genes that had hypothesized relationships with personality and were known to be expressed in the brain in other species, including humans and domestic dogs. The researchers also concluded that there are at least five factors that comprise elephant personality: dominance, neuroticism, agreeableness, curiosity, and impulsiveness. "This is the first report," they write, "of an association between a genetic polymorphism and personality in elephants."

Unfortunately, the personality assessments were problematic. The researchers noted that the "elephants in Western zoos were rated as being more curious, and less dominant, impulsive, and nervous than elephants in Japanese zoos." Could Western elephants really be more curious than their counterparts living in Japanese zoos?

What could explain this apparent group difference? It could be that the distinction is legitimate. Perhaps, as a result of random chance, the more curious elephants wound up in American and Canadian zoos, and the more dominant, impulsive, and nervous ones were placed in Japanese zoos. It is more likely, however, that the assessments measure the keepers' perceptions of elephant behavior rather than elephant behavior per se. The differences that the researchers found might therefore reflect differences among Japanese and Western keepers! It's the problem of David and Dana, all over again.

Recall that personality should be independent of environment. That is, if you took an elephant from an American zoo and placed it into a Japanese zoo, its basic personality would remain unchanged. This study suggests that it's personality assessment, however, might not. And that's a problem. To avoid the issue, the researchers looked at the gene-personality correlations only among the elephants from Japanese zoos. And so seventy-five elephants with both personality and genetic data became forty-five. Those forty-five were really one group of seventeen Asian elephants and a second group of twenty-eight African elephants. Here's why.

The elephants in the study came from two different genera: Loxodonta and Elaphas. Genera is the plural of genus, which is the level of taxonomic classification above species. These two elephant species share designation at the family level, Elephantidae. By comparison, the family Hominidae includes four genera: Pan (chimpanzees and bonobos), Gorilla (gorillas), Homo (humans), and Pongo (orangutans). The African and Asian elephants diverged from their common ancestor approximately seventeen million years ago. By comparison, chimpanzees and humans diverged approximately six million years ago. In other words, the two species of elephants that were included in this study have been evolving separately nearly three times as long as have humans and chimpanzees. This is why the researchers quite rightly separated their analyses of Asian and African elephants: meaning that they had two small elephant groups, rather than one larger (but still relatively small) group.

University of British Colombia psychologists Joseph Henrich, Steven J. Heine, and Ara Norenzayan argued, in a 2010 paper published in the journal Behavioral and Brain Sciences, that psychological scientists had erred in deriving conclusions about human nature from studies relying only on individuals from Western, Educated, Industrialized, Rich, and Democratic (WEIRD) societies. They found, when comparing Western and non-Western populations, that there were differences in abilities ranging from visual perception to moral judgement. Even estimates of the heritability of intelligence varied according to culture. (See The WEIRD Evolution of Human Psychology by Eric M. Johnson)

Might this study of elephant personality, which may truly be measuring keeper values, be suffering from a similar problem? Not only was the personality measurement possibly flawed, as it may have been susceptible to the cultural biases of elephant keepers (like the David and Dana infant personality study), but the researchers then tossed out all the data from the Western keepers (an inverse of the WEIRD problem)!

The researchers write, "this study is the first examination of genetic components affecting individual differences of personality in captive elephants." But if the personality questionnaire was truly a measurement of culturally-derived values among keepers instead of personality differences among elephants, then the observed correlation between neuroticism and the ASH1 gene in Asian elephants would be rendered meaningless. The lessons learned from decades of research in human psychology ought to inform that way we design our studies of non-human animals.

Yasui S, Konno A, Tanaka M, Idani G, Ludwig A, Lieckfeldt D, & Inoue-Murayama M (2012). Personality Assessment and Its Association With Genetic Factors in Captive Asian and African Elephants. Zoo biology PMID: 22996044


Feeling Snappy? Measuring Personality in Hermit Crabs

Images: African elephant photo copyright the author, taken at the San Diego Zoo November 20, 2011.

Update: Psychologist Dave Nussbaum points out on twitter that personality may not be as stable across environments as personality theorists might argue. It is indeed possible that there are sufficient differences between Western and Japanese zoos, such as in elephant husbandry techniques or exhibit design, that could result in the sorts of group differences observed in this study. I'd offer that this possibility doesn't eliminate the possibility of keeper bias. While the researchers did collect data on housing style and management procedures, it is unclear whether there were Western/Japanese differences for those variables, as the data from Western elephants were not included in the analyses.