Starting from the Beginning: Evolutionary and Developmental Origins of Human Knowledge

What are the cognitive and neural systems that allow us to build buildings, play checkers, do multivariate statistics, receive DVDs by mail, follow Dr. Isis's pesto recipe, or navigate the tangled LA freeways?

You may ask: what can studying children and non-human animals tell us about the complexity of the human experience? Only educated human adults engage with formal mathematics, cooking, or map reading. Right?

This is a reasonable question to ask. But when human adults show complex, possibly culture-specific skills, they emerge from a set of psychological (and thus neural) mechanisms which have two properties:
(1) they evolved early in the timecourse of evolution and are shared with other animals, and,
(2) they emerge early in human development, and can be found in infants and children, as well as adults.

These more basic psychological/neural mechanisms are called core knowledge systems. I will propose (and I hope to convince you, dear reader) that in order to understand the human mind, it is important to investigate both the core knowledge systems - the evolutionary building blocks of the mind - as well as the processes by which these systems are combined to give rise to the richness of human culture we see today.

Four main properties of core knowledge systems emerge from decades of research:
(1) Domain-specificity. Each system represents a particular kind of concept or percept, and nothing else. There are systems for objects, conspecifics (others of the same species), places in a scene, or numerosities, for example. The numerosity system does not operate over conspecifics (except for when counting them).

(2) Task-specificity. The cognitive representations from each system address specific questions about the world. For example, the face recognition system answers questions like, "who is this?" The spatial orientation system answers questions like, "where am I?"

(3) (Relative) Encapsulation. Each system uses a subset of the information received from the sensory input systems, and sends information only to a subset of the animal's output systems. With some exceptions, these systems don't really interact with each other.

(4) Automaticity. These systems operate independently and are unaffected by explicitly held beliefs or goals.

How can we investigate the evolution and architecture of core knowledge?

Comparative Approaches
Marc Hauser and Elizabeth Spelke provide the following thought experiment:

You are an alien scientist from another planet and you descend to earth to find only one species: Homo sapiens. You are interested in why this species looks and behaves as it does, as well as why it has the kind of social organization observed. How might you find answers to these functional problems? Since you are an atheist alien, you sweep aside explanations of divine cause. What you turn to instead is an exploration of evolutionary design, of trying to understand the relationship between the trait's engineering and the kind of problem it appears to solve. Having read Darwin, you might conclude that only natural selection can account for a trait with complex design features - an assemblage of parts with a nonrandom probability of developing together. But there is a deep problem with this approach. Although you might find out something about current function, you will never know whether this represents its original function; nor will you understand whether the now extinct inhabitants of earth, including all of the direct and indirect predecessors of Homo sapiens, had similar qualities; and neither will you understand whether the observed trait represents the only solution to the problem at hand. Finding fossils of other species will help explain anatomical traits, but will only lead to speculation about behavior, psychological representation, and neural instantiation. You, the alien scientist, are stuck.

Fortunately, we do not face the problem of the alien scientist. We are surrounded with myriad other species - some closely related, others more distantly so.

Three questions necessitate a comparative evolutionary approach (or, minimally, are enriched by such an approach):
(1) Is a given trait unique to humans?
(2) Does the acquisition of a given trait depend on uniquely human abilities?
(3) What functional problem does a given trait solve, and did it evolve for this particular function?

That the first question necessitates a comparative approach should be obvious. If comparative data indicate that even only one other species possesses the trait in question, then the question shifts a bit, and we have to determine whether the trait is homologous (depending on the same mechanisms), or homoplastic (depending on distinct mechanisms that presumably evolved independently). How can we distinguish homology from homoplasy? We look for signatures, or common features. For example, face processing in humans shows behavioral signatures (e.g. degradation when faces are inverted) and neural signatures (localized cortical activations). Those same features have been found in various monkey species that have been tested in face processing tasks, and this provides one piece of evidence for homology.

The second question asks if one cognitive capacity depends on the existence of another cognitive capacity. Does theory of mind (the ability to infer the mental states of others) depend on language acquisition? If so, you would not expect to find evidence for theory of mind in non-human primates, who lack several of the computational abilities that give rise to language. That we do find evidence for theory of mind in other species suggests that the core capacity for inferring the mental states of others emerged in evolution independently of language.

The third question distinguishes among the original function of a trait and the way it is currently used. Language, for example, allows us to recombine a finite set of elements in essentially infinite patterns to create meaning. Did this capacity evolve to facilitate communication, or for some other purpose? Assume that chimpanzees, for example, do not show evidence of this mechanism in their communication, but DO exhibit this mechanism for arithmetic computation. This might suggest that this ability evolved for number, and was then "re-purposed" by humans for communication. Of course, it is also possible that this capacity evolved independently in chimpanzees and in humans, but this seems less likely given the relatedness of our two species.

A fourth essential feature of animal studies is the ability to control the rearing environment (and therefore, the experience) of a developing animal. Hubel and Wiesel's landmark studies, for example, investigated the effects of binocular deprivation on the development of binocular vision in cats and monkeys. Similarly, Gibson and Walk's dark-rearing studies contributed a lot to the understanding of visually-guided locomotion. Do certain abilities require certain environmental input, or do they emerge spontaneously, independent of experience? If so, we can reasonably infer that such an ability is innate.

Developmental Approaches
Once again, consider our alien scientist. Upon first glance, human infants seem pretty incompetent. At birth, they have terribly low visual acuity, are unable to move with any coordination, accomplish nothing by their actions, and say nothing. In a few short years, children learn a great deal about the world. But maybe infants come into the world with a set of really powerful cognitive structures, and because of certain developmental constraints (e.g. the inability of the neck muscles to support the weight of the head), are unable to properly display or use them. The key then, is to figure out the proper testing conditions to measure these capacities.

Prior research has indicated that infants discriminate speech sounds and separate the speech stream into syllables and words, perceive depth and use that information to guide motion, and categorize the visual scene into objects. More recent research indicates that infants can distinguish animate from inanimate objects (i.e. agents and objects), reason about numerosities, and distinguish among intentional and accidental actions.

Babies aren't so incompetent, after all! The logic of comparative animal studies applies here as well. If infants, children, and adults show the same signatures for a given ability, then you can infer developmental continuity. Then one can investigate further, perhaps using neuroimaging, to see whether common neural signatures underlie the same processes in infants, children, and adults. The next step might be to compare neural systems used in humans throughout development for a given task, and compare with the neural systems used by non-human primates. And finally, one can map those findings onto molecular genetic commonalities and differencies. Evidence for the effectiveness of this paradigm can be seen in the 2002 findings that the FOXP2 gene is recently evolved and unique to our species, and subserves some aspects of language (more recent evidence suggests that FOXP2 is not be unique to humans, though the changes in FOXP2 that may underlie some of our language abilities *is* recent).

Whew! That was probably enough for today. Have at it in the comments!

Hauser, Marc D., & Spelke, Elizabeth (2004). Evolutionary and Developmental Foundations of Human Knowledge The Cognitive Neurosciences III (Ed. M. Gazzaniga)