Research confirms that there is a general intelligence factor. it predicts

In humans cognitive abilities such as navigating through space, understanding written language and number skills correlate positively; a person who is above average at one task is likely to be good at others (Deary et al., 2010, Deary, 2013). Hundreds of empirical phenotypic studies show that the structure of human abilities can be represented as a hierarchy with observed manifest measures or tests (such as verbal comprehension or arithmetic) at the bottom level, latent group factors (such as spatial or verbal skills) at the second level and a third factor at the apex (Carroll, 1993). This third factor, called g or Spearman's g after its discoverer Charles Spearman (Spearman, 1927), is a major focus of psychometric studies in the human behavioural sciences (Jensen, 1998, Johnson et al., 2004, Spinath et al., 2003).

Quantitative genetic methods developed in the 1970s and applied to data from adoption and twin studies have established the existence of genetic g; that is, abilities are correlated at the genetic as well as the phenotypic level (Bouchard and McGue, 1981, Deary et al., 2006, Loehlin et al., 1997, Pedersen et al., 1992). More recently, evidence from molecular genetic studies using DNA from large samples of unrelated people show that g is highly polygenic (Davies et al., 2011). Research on g is motivated partly because it is phenotypically associated with many important life outcomes including health (Batty et al., 2007, Luciano et al., 2010, Mõttus et al., 2013, Schou et al., 2012), physical attractiveness (Langlois et al., 2000, Zebrowitz et al., 2002), brain resilience (Santarnecchi, Rossi, & Rossi, 2015), and life-expectancy (Batty et al., 2009, Batty et al., 2007, Whalley and Deary, 2001). The phrase cognitive epidemiology was coined to characterise research into the association between measured intelligence and traits such as health and life-expectancy in people (Deary & Der, 2005). It would be useful to learn whether the pattern of findings linking higher g with better health outcomes (Gottfredson, 2004) is particular to people or common among animals. Links between intelligence and health in non human animals would be especially interesting to probe because other animals neither smoke nor drink alcohol (habits that are lifestyle confounders in human studies). But as the legendary recipe prescribes, ‘first catch your hare’; in this case, evidence concerning the structure of cognitive abilities in other species. This ‘hare’ is an essential first step in probing a link between intelligence and health in other species.

There is some evidence of g in non human animals (reviewed in Chabris, 2007, Galsworthy et al., 2013, Matzel et al., 2013). Yet evidence of the distribution, structure (phenotypic and genetic correlations among cognitive abilities), and the consequences of those differences in other species is exiguous: relatively few studies on general intelligence have been conducted in non human animals since 1920 (one review comprised 21 studies (Chabris, 2007), another comprised 24 studies (Galsworthy et al., 2013)). In order to test whether cognitive abilities are correlated or not, individual-level data on task performance need to be collected, in a sample of reasonable size. This has been done in mice (Galsworthy et al., 2002, Locurto et al., 2002, Matzel et al., 2003, Wass et al., 2012), where a g factor was found, and in chimpanzees (Banerjee et al., 2009, Herrmann and Call, 2012, Hopkins et al., 2014) where a g factor was found in two out of three studies.

We tested the structure of measured cognitive abilities in dogs. Dogs and dog breeds are good models for within- and between-species spectra of cognitive abilities. The reasons are plural. Dogs are tractable; they enjoy interacting with people and can visit testing facilities, while living in their own homes. 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, 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, Shipman, 2010).

Our underlying assumption was that cognitive abilities would vary among dogs. This is implied by existing data in the animal behaviour literature but variance is rarely the focus of the work. For example, many animal cognition studies are framed as ‘can species X do the Y task?’ yet the results usually include animals that did, and did not, pass the test. Behavioural variability is the rule not the exception; since variance supplies evolution with its traction, it is a worthwhile object of study.

The present empirical study owes an intellectual debt to the work of John Paul Scott and John L Fuller (Scott & Fuller, 1965). 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, Scott and Fuller, 1965), a quantity-discrimination task (Bonanni et al., 2011, Macpherson and Roberts, 2013, Prato-Previde et al., 2008, 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, 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.

Cited by (36)

• The contribution of executive functions to sex differences in animal cognition

  2022, Neuroscience and Biobehavioral Reviews

  Show abstractNavigate Down

  Cognitive sex differences have been reported in several vertebrate species, mostly in spatial abilities. Here, I review evidence of sex differences in a family of general cognitive functions that control behaviour and cognition, i.e., executive functions such as cognitive flexibility and inhibitory control. Most of this evidence derives from studies in teleost fish. However, analysis of literature from other fields (e.g., biomedicine, genetic, ecology) concerning mammals and birds reveals that more than 40% of species investigated exhibit sex differences in executive functions. Among species, the direction and magnitude of these sex differences vary greatly, even within the same family, suggesting sex-specific selection due to species’ reproductive systems and reproductive roles of males and females. Evidence also suggests that sex differences in executive functions might provide males and females highly differentiated cognitive phenotypes. To understand the evolution of cognitive sex differences in vertebrates, future research should consider executive functions.

• How spontaneous is spontaneous quantity discrimination in companion dogs?

  2022, Applied Animal Behaviour Science

  Show abstractNavigate Down

  Although quantity discrimination and the factors affecting it have been widely studied in the domestic dog using a variety of paradigms, little attention has been given to the possible effects of the used paradigm itself. In the present study, we employed a paradigm in which naïve companion dogs were repeatedly presented with a free choice between two quantities of food (2vs.4, more difficult comparison, and 1 vs.8, simpler comparison). Dogs did not undergo any previous training and could freely choose to feed from either plate. After the choice was made, the second plate was withdrawn without letting the dog to eat its content. We hypothesized that a preference for the larger quantity of food may emerge as a consequence of experiencing the experimental procedure, and may not be expressed by the dogs on their first choice (hypothesis 1). If so, rewards experienced in the first trial may affect behaviour on subsequent trials (hypothesis 2). Data were analysed using generalized linear mixed models. Both hypothesis 1 (general performance P < 0.0001, i.e., above chance level; first choice P = 0.06, i.e., chance level) and 2 (P = 0.001) were confirmed by the experimental results. The difficulty of the numerical comparison and the neuter status of the dogs had a significant effect on the overall performance, but not on the first choice nor on the likelihood of redirecting after the first trial. The results suggest that domestic dogs are highly sensitive to the results of their experience and adjust their behaviour accordingly, even after one single event. Future studies may help to disentangle the role of the food seen vs. eaten by the dog and the possible role of unintentional human feedback after the first choice.

• Study of variability of cognitive performance in captive fallow deer (Dama dama) through g and c factors

  2022, Journal of Veterinary Behavior

  Show abstractNavigate Down

  Psychometric studies reveal a variability that ranges between 30 and 75 % for different cognitive processes, both in humans and non-human animals. This variability has been ascribed to a “general intelligence” factor (g). Current studies implemented in non-human animals to identify individual differences in cognitive skills are still far from those applied in humans. We tested for cognitive performance in a fallow deer herd using a test battery of eight tasks based on measures for cognitive development in untamed confined wild animals: motor control, spatial memory, colour and tone discrimination and inversion, inhibitory control and symbol discrimination tests. g factor explained 60 % of the variability in cognitive performance. Significant correlations, both positive and negative, were found between the different cognitive processes studied. Likewise, the cognitive performance of the group was affected by a “collective intelligence” factor (c), inferred through variables describing the hierarchical structure of the studied population but which, however does not affect individual intelligence quotient of the members of the group.

• Differential and experimental approaches to studying intelligence in humans and non-human animals

  2020, Learning and Motivation

  Show abstractNavigate Down

  Why do some individuals learn more quickly than others, or perform better in complex cognitive tasks? In this article, we describe how differential and experimental research methods can be used to study intelligence in humans and non-human animals. More than one hundred years ago, discovered a general factor underpinning performance across cognitive domains in humans. Shortly thereafter, discovered positive correlations between cognitive performance measures in the albino rat. Today, research continues to shed light on the underpinnings of the positive manifold observed among ability measures. In this review, we focus on the relationship between cognitive performance and attention control: the domain-general ability to maintain focus on task-relevant information while preventing attentional capture by task-irrelevant thoughts and events. Recent work from our laboratory has revealed that individual differences in attention control can largely explain the positive associations between broad cognitive abilities such as working memory capacity and fluid intelligence. In research on mice, attention control has been closely linked to a general ability factor reflecting route learning and problem solving. Taken together, both lines of research suggest that individual differences in attention control underpin performance in a variety of complex cognitive tasks, helping to explain why measures of cognitive ability correlate positively. Efforts to find confirmatory and disconfirmatory evidence across species stands to improve not only our understanding of attention control, but cognition in general.

• Sleep and intelligence: critical review and future directions

  2020, Current Opinion in Behavioral Sciences

  Show abstractNavigate Down

  General cognitive ability — or intelligence — is a key psychological phenotype. Individual differences in intelligence may either cause or be a consequence of individual differences in the macrostructure of sleep, such as timing or duration. Furthermore, biological measures of sleep, especially highly trait-like sleep EEG oscillations may provide insights about the biological underpinnings of intelligence. Here we review the current state of research on the association between sleep measures and intelligence. We concluded that the macrostructure of sleep has a small but consistent correlation with intelligence, which is possibly moderated by age. Sleep spindle amplitude and possibly other sleep EEG measures are biomarkers of intelligence. We close by discussing methodological pitfalls of the field, and give recommendations for future directions.

• The exceptions that prove the rule? Spontaneous helping behaviour towards humans in some domestic dogs

  2020, Applied Animal Behaviour Science

  Show abstractNavigate Down

  Domestic dogs (Canis familiaris) are skilled at reading and correctly responding to human communicative gestures to locate hidden food. Whether they, like chimpanzees, will understand requests for help in retrieving a fallen object, is not known. The aim of this study was to examine whether dogs show spontaneous helping behaviour towards a human experimenter that tries to obtain an object that is out of reach. The object at stake either “accidentally” fell on the floor, or was thrown on the floor by either a familiar (owner) or unfamiliar human. In order to get a better understanding of individual differences between helping and non-helping dogs, the behaviour of all dogs was observed by means of continuous focal animal sampling and scored by means of an ethogram. Personality traits were measured by letting owners rate their dogs on 50 personality adjectives using a 7-point Likert scale. The results demonstrate that six out of 51 dogs showed helping behaviour and did so more in the accidental (experimental) condition, than when the object was thrown on the floor on purpose (control) condition (P = 0.001). Dogs in general wagged their tail more (P = 0.009) and looked less often towards the test leader (P < 0.001) in the experimental condition compared to the control condition, suggesting that they experienced more arousal whenever humans were in need of help. In addition, a principal component analysis indicated to retain 41 adjectives which revealed five personality factors, in line with previous research, that accounted for 60.7 % of the total variance. However, the six exceptional dogs had no outstanding personality traits and were of different breeds suggesting that this did not explain the differences in helping behaviour. We conclude that dogs appear motivated and willing to help humans, but that the majority does not understand the source of the problem or how to assist. We discuss this result in light of the previously reported social skills of dogs and nonhuman primates.

Recommended articles (6)

• Research article

  IQ and socio-economic development across local authorities of the UK

  Intelligence, Volume 55, 2016, pp. 90-94

  Show abstractNavigate Down

  Cross-regional correlations between average IQ and socio-economic development have been reported for many different countries. This paper analyses data on average IQ and a range of socio-economic variables at the local authority level in the UK. Local authorities are administrative bodies in local government; there are over 400 in the UK, and they contain anywhere from tens of thousands to more than a million people. The paper finds that local authority IQ is positively related to indicators of health, socio-economic status and tertiary industrial activity; and is negatively related to indicators of disability, unemployment and single parenthood. A general socio-economic factor is correlated with local authority IQ at r = .56. This correlation increases to r = .65 when correcting for measurement error in the estimates of IQ.

• Research article

  Does intelligence affect health care expenditure? Evidence from a cross-country analysis

  Intelligence, Volume 55, 2016, pp. 86-89

  Show abstractNavigate Down

  Most of the literature on the determinants of health expenditure is focused on the relationship between health care expenditure and income. We investigate the interactive effect of income and intelligence on health expenditure, using data from 172 countries, over the period from 2009 to 2013. The results show that the income elasticity of health expenditure is greater than 1 but only if a nation's level of IQ score beyond a certain threshold (about 95), while for low-IQ nations, health care is a necessity. Moreover, various robustness checks largely support the robustness of our main findings.

• Research article

  A neural analogue of the worst performance rule: Insights from single-trial event-related potentials

  Intelligence, Volume 55, 2016, pp. 95-103

  Show abstractNavigate Down

  The worst performance rule is the tendency for participants' slowest reaction times to correlate more with psychometric intelligence than their faster reaction times. Reaction times, however, are influenced by the duration of multiple perceptual, attentional, and motor sub-processes, and it is unclear whether the same pattern exists in these sub-processes as well. We used single-trial event-related potentials to identify whether a worst performance rule pattern could be found in stimulus and response-locked P3b latency distributions and scores on a test of non-verbal psychometric intelligence.

  Fifty participants carried out a set of working memory oddball tasks, while electroencephalographic data were collected, and the British Version of the Intelligence Structure Test, in a separate session. Single-trial P3b latencies were identified in stimulus and response-locked data and a novel quantile bootstrapping method was used to identify which quantiles of the P3b latency distributions correlated most with test scores.

  In stimulus-locked data, correlations between quantile mean and test scores became more negative with increasing quantile, showing clear evidence of a worst performance rule pattern. In response-locked data, low scorers showed more extreme latencies in both tails of the distribution. However we did not observe a worst performance rule in behavioural data. These data suggest that psychometric intelligence is also associated with response-related processes, which may also contribute to the association between psychometric intelligence and reaction time variability.

• Research article

  Residual group-level factor associations: Possibly negative implications for the mutualism theory of general intelligence

  Intelligence, Volume 55, 2016, pp. 69-78

  Show abstractNavigate Down

  The mutualism theory of general intelligence (g) posits that the positive manifold arises because of mutually beneficial interactions between originally orthogonal cognitive abilities, rather than because of a genuine general construct. In a recent investigation, Gignac (2014) reported that the strength of g was largely constant from the ages of 2.5 to 90 years, which was interpreted as an indirect failure to confirm the mutualism theory of g. In this investigation, a second indirect test of the mutualism theory of g was performed. Specifically, it was hypothesized that, if the extended mutualism theory of g is plausible, then there should be some consistent, positive associations between group-level factors, controlling for the effects of a general factor. To examine this possibility, the associations between cognitive ability group-level factors were estimated across a series of seven relatively large and relatively representative samples of intelligence battery data. Next, seven single-factor models were estimated against the seven group-level inter-correlation matrices. The corresponding single-factor residual correlation matrices were observed to yield an approximately equal number of positive and negative residual correlations and an overall mean of zero. Furthermore, the only moderately consistent residual effect was a negative association between crystallised intelligence and processing speed. Although only an indirect test, the results are interpreted to be more supportive of g factor theory than mutualism.

• Research article

  What counts for dogs (Canis lupus familiaris) in a quantity discrimination task?

  Behavioural Processes, Volume 122, 2016, pp. 90-97

  Show abstractNavigate Down

  Numerous studies have reported that animals reliably discriminate quantities of more or less food. However, little attention has been given to the relative salience of numerosity compared to the total amount of food when animals are making their choices. Here we investigated this issue in dogs. Dogs were given choices between two quantities of food items in three different conditions. In the Congruent condition, the total amount of food co-varied with the number of food items; in the Incongruent condition the total amount was pitted against the numerosity; and in the Controlled condition the total amount between the sets was equal. Results show that dogs based their choice on the total amount of edible food rather than on the number of food items, suggesting that, in food choice tasks, amount counts more than number. The presence of the largest individual item in a set did not bias dogs’ choices. A control test excluded the possibility that dogs based their choices on olfactory cues alone.

• Research article

  Can dogs count?

  Learning and Motivation, Volume 44, Issue 4, 2013, pp. 241-251

  Show abstractNavigate Down

  Numerical competencies have been thoroughly examined in several species, yet relatively few studies have examined such processes in the domestic dog. In an initial experiment, procedures from numerical studies of chimpanzees (, ) were adapted for use with 27 domestic dogs. Subjects in these experiments watched as different quantities of food were sequentially dropped into each of two bowls. The subjects were then allowed to select and consume the contents of one of the bowls. Although dogs excelled in a 1 vs 0 condition, their performance failed to significantly surpass chance across all other ratios. In a second experiment with a single subject (a rough collie named Sedona), the procedure was revised so that non-food stimuli were presented simultaneously to the dog on two magnet boards. If Sedona chose the board with the majority of the items, she was rewarded with a piece of food hidden underneath the board. If she made an incorrect choice, she received no reinforcement. Interestingly, Sedona's performance far exceeded that of the dogs in Experiment 1. Implications of these findings for the study of domestic dogs are discussed.