Defining-Features Representations Restored.

When my cousin, Samuel, was two or three years old, he thought that all Daddies were Indian, and all Mommies were Chinese. At first glance, this might seem like a really strange idea to have, but it’s really not all that surprising. The thing is, my father is Indian, and on his side of the family I have two uncles, one of whom is my cousin’s father. All three of them (my father and both uncles) married Chinese women and so all my cousins are “Chindians”, as we’re called in Malaysia. Thus, in the early years of his life, Samuel would have come into contact with mainly Indian Daddies and Chinese Mommies, which led to his unusual conclusion.

Samuel wasn’t only exposed to relatives growing up – he met other Mommies and Daddies too. Therefore he would’ve met other Daddies and Mommies who weren’t Indian and Chinese respectively. However, he spent most of his time within his nuclear and extended family, so the “being Indian” feature was a high cue validity for Daddies, as was the “being Chinese” feature for Mommies. As he grew up, he began to learn what exactly makes a Mommy and a Daddy. He was taught that anyone who had children was a Mommy is she was a girl, and a Daddy if he was a boy. Despite the fact that he was still mainly exposed to his nuclear and extended family, he now knew that race does not have a causal relationship with Mommy- and Daddyhood. His categories of Daddies and Mommies were both broadened and limited by this new information that he now had, and were also more accurate.

My point to this analogy is that the three theories of conceptual representations need not be seen as mutually exclusive, but can instead be combined to form a single continuous and dynamic theory of how children’s conceptual representations develop. While probabilistic representations and theory-based representations are seen to complement each other, the defining-features representations theory is often deemed too basic and therefore obsolete. I beg to differ – I think the defining-features theory might very well be the culmination of children’s conceptual representations.

Going back to my analogy, Samuel first used probabilistic representations to form his categories of Daddies and Mommies. He could not have used defining features, as he was exposed to parents of other races, neither could he have used theory-based representations because he did not know what caused a person to be a Mommy or a Daddy. The second step in Samuel’s representations of Mommies and Daddies was theory-based – he learnt what “made” Mommies and Daddies. Based on what he learnt, he could then form defining features of Mommies and Daddies – Mommies were girls who had children, and Daddies were boys who had children.

Mommies and Daddies aside, this process of forming conceptual representations can be seen in many other instances. Under the Naïve Biology theory, for example, children first start off by categorizing living and non-living objects based on motion. Most of the things they are exposed to early in life that aren’t alive don’t move – their teddy bear, their blanket, and their bottle, for instance. They would almost definitely have some experience with plants, and might even be told that the plant is alive and can die. However, as the vast majority of inanimate objects they experience are non-living, non-motion is a high cue validity for non-living objects and thus their categories of living and non-living objects are formed.

Later on, they are taught either formally or informally about what exactly makes living things “alive”. They learn about respiration, reproduction and nutrition, to name a few, that is causally related with “alive-ness”. Using these theory-based representations, they then modify and possibly do away with the probabilistic representations they had, if need be, and identify the defining features that categorize living and non-living things. This categorization would be more accurate than the one formed using probabilistic representations.

One of the weaknesses that have been identified for defining-features representations is that there are some concepts that do not have defining features – I don’t think this is true because these defining features are there; it’s just a matter of learning what they are. Probabilistic representations, on the other hand, have the weakness of possibly being and remaining inaccurate if not coupled with theory-based representations. The two combined then result in defining-features representations, and so these two processes are in fact the way in which we learn what the defining features for a concept are.

In summary, with limited knowledge and experience, children use cue validities when faced with novel objects and experiences. When they gain more knowledge about those objects and experiences, they might be able to form theory-based representations which they can then use to form defining-features representations. If these defining-features representations coincide with their probabilistic representations, all the better, but if they don’t, then the defining-features representations take precedence. This proves the relevance, and indeed the superiority, of defining-features representations over probabilistic and theory-based representations.