There are no "natural kinds of basic behavioural units of culture equivalent to genes, cells, and organisms. Therefore each major discipline- anthropology, sociology, political science, and so forth- has been required to develop its own conceptual base and language. ("nomic isolation." )

The discovery of natural units in culture would represent a key advance in the social sciences. Most appear to believe that such units either do not exist or, if they do exist, cannot be derived by any means currently available.

However, there is some reason to believe that natural units do exist and are built on the natural units of semantic memory.

Semantic memory comprises words and symbolic manipulation.

Episodic memory entails running through sequences of visual and other sensory experience. It tends to organize impressions into discrete clusters. Experimental studies have revealed that mental cuts are made around objects or abstractions that have the most attributes in common. Hence although categories such as "tree," "dog," and "house" are collections of objects that share a relatively large number of stimuli most easily processed by the brain.

Children move easily into the semantic mode of memory formation, performing equally well on objects or collections of objects. They organize certain identifying stimuli into sets (such as "cookies" versus "cakes" and "chairs" versus "stools") that are almost as sharply demarcated as the separate objects themselves.

The brain process these sets further by compounding the clusters hierarchically into larger assemblages that possess a discrete, interchangeable form.

The units of semantic memory, which are experienced as obects or abstractions, are appropriately called nodes.

There are at least three levels of nodes.

• Concepts, the most elementary clusters, are expressed by words or phrases (such as "dogs" and "hunt").
• Propositions are signalled by phrases, clauses, or sentences expressing objects and relationships ("dogs hunt").
• Schemata are signalled by sentences and larger units of text (the "technique of hunting with dogs"). 

Node-link structures steadily enlarge in size and complexity in the growing child, and the main steps in the growth correspond at least roughly to the stages of mental development. The stages are not mere accidents of personal growth but general processes that show some regularity across cultures.

It is important for the relation of biology to culture that the semantic mechanisms of culture formation are more robust and consistent than the final products they generate. For each concept the brain tends to select a prototype that constitutes the standard, such as a particular wavelength and intensity to form the idealized colour red or a particular body shape and size to form the typical "dog."

Given an array of similar variants, the mind can deduce a standard near the average of the variants and use it as the prototype even without having perceived any example of it directly. The most important result for gene-culture coevolution is that the divisions are created and labelled, even when the stimuli being processed vary continuously.

In short, the mind automatically imposes a semidiscrete, hierarchical order upon the world.

Most of the concepts that make up the basic units of semantic memory are arbitrary and arise from the particularities of cultural history. Nevertheless there is a tendency for those belonging to at least a few categories to occur consistently across cultures.

Such categories include elementary geometric forms (square, circle, equilateral triangle), the facial expressions of six basic emotions (happiness, sadness, anger, fear, surprise, disgust), and the basic colors (red, yellow, green, blue).

The level of the node of semantic memory, whether concept (the most elementary recognizable unit), proposition, or schema, determines the complexity of the generated behaviour or artifact maintained in the culture.

For example, the differentiation of letters or ideographs is at the level of the concept, the initial verbal reaction to a stranger is a proposition, and the expression of an incest taboo is a schema.

If this model of semantic memory holds, new discoveries refining the hierarchy of memory nodes can be expected to advance the identification of culture units, or "culturgens," in the same manner that advances in cellular chemistry have improved our comprehension of the gene and studies of population structure have refined our understanding of biological species.

Although a direct correspondence between nodes and generative units of culture appears feasible at lower levels of organization, there is no reason to expect the more complex constructions of culture to be mapped onto semantic nodes in a one-to-one fashion. Marriage ceremonies and temple architecture, for example, are the outcomes of numerous interlocking behaviours that result from cognitive activity with multiple culturgens. These in turn vary according to the particularities of local history. Nevertheless, each can be realistically interpreted as the outcome of cognitive development, which is attained principally through the assembly of node-link networks.