Categories and Inductive Inferences
Source: Laboratories of Nicholaus Noles and Judith Danovitch—University of Louisville
It might be possible for the human brain to keep track of each i…
Recruit healthy 5-, 7-, and 9-year-old children with normal hearing and vision and no history of developmental disorders. For the purposes of this demonstration, only one child is tested. Larger sample sizes (as in Frank Keil's transformation study1) are recommended when conducting any experiments.
1. Data collection
- Gather the necessary materials.
- Design eight vignettes describing the transformation of animals from one kind to another via surgery.
- Natural kind vignette example: "The doctors took a raccoon and shaved away some of its fur. They dyed the rest of the fur black. Then, they bleached a single white stripe down the center of its back. They surgically added a sack of super smelly stuff into its body, just like a skunk has. When they were all done, the animal looked like this. After the operation, was this a skunk or a raccoon?"
- Design eight vignettes describing the transformation of artifacts from one kind to another via physical alteration.
- Artifact vignette example: "The doctors took a coffeepot that looked like this. They sawed off the handle, sealed the top, took off the top knob, sealed the spout, and sawed it off. They also sawed off the base and attached a flat piece of metal. They attached a little stick, cut a window in it, and filled the metal container with bird food. When they were done, it looked like this. After the operation, was this a coffeepot or a birdfeeder?"
- Create cards with images of each item before and after transformation.
- Design eight vignettes describing the transformation of animals from one kind to another via surgery.
- Preamble
- Present the child with a preamble introducing doctors and describing operations and surgery.
- Test
- Present the child with each of the 16 vignettes in a random order.
- If the child tries to give an ambiguous response (e.g., "It's both" or "It's a skuncoon"), ask them to pick one of the two highlighted responses.
- Question the child in a free-form manner in order to understand what aspects of the vignette guided their responses. For example, the experimenter might ask questions like: "Why is it a birdfeeder?" or "If the animal looks like a skunk and smells like a skunk, why did you say it was a raccoon?" This questioning helps to determine what features or principles the child uses to guide their intuitions.
- Record the child's responses, and then transcribe them for future analysis.
2. Analysis
- Two independent coders read the child's transcriptions and code them on a scale from 1 to 3 for each item. This score is an index of the child's resistance to categorical change.
- A score of 1 on an item means a child shifts their intuitions to follow the transformation operation. These responses focus on characteristic features.
- Example: "It's stinky and stripy like a skunk, so it's a skunk."
- A score of 2 on an item means a child's responses are indecisive, indicating that they are unsure if the transformation results in a change in the item's kind.
- Example: "It's a rac-unk or a skoon. I don't know."
- A score of 3 on an item means that a child resists shifting their intuitions following transformation operations. These responses contrast with observed features.
- Example: "It doesn't matter what it looks like on the outside; it's still a raccoon."
- A score of 1 on an item means a child shifts their intuitions to follow the transformation operation. These responses focus on characteristic features.
- Use an analysis of variance to determine if there are differences between the three age groups and two types of items presented to the children.
To organize the world and efficiently use cognitive resources, children learn to place objects, people, and locations into categories.
In a broad sense, a category?like fruits?is a mental representation of a collection of real things that share similar features, such as the presence of seeds.
In this instance, the characteristic of having seeds is a "hidden" or internal feature?one not easily observed?that can be used to link items to form a category.
However, sometimes the feature?or features?on which a category is based can be readily perceived. For example, if a child sees an animal that has four legs, a tail, and barks, they?ll likely categorize it as a dog.
Importantly, children can take what they know about a single item belonging to a category, like their pet dog, and use that information to identify and make inductive inferences?educated guesses?about unfamiliar category members.
For example, if a child knows that their dog wags its tail when it?s happy, then they can guess that a new tail-wagging dog is also happy.
Based on the pioneering work of Frank Keil, this video explores how to investigate the criteria 5- to 9-year-olds use to categorize objects by designing stimuli that cross categorical boundaries, and explains how to perform such a transformational study and interpret categorization data. We will also look at how psychologists use this technique in other applications, for example to explore how children make inferences.
In this experiment, researchers present 5-, 7-, and 9-year-old children with 16 stories in which items are physically altered so that they resemble something else.
Eight of these stories center around living animals or naturally occurring objects, like zebras or diamonds?collectively called natural kinds?and eight deal with man-made, non-living items referred to as artifacts, such as cups or pots.
For instance, in a natural kind-focused transformation story, children are first shown a "before" image of a naturally occurring item or animal, such as a raccoon.
They?re then told that doctors surgically transformed this thing, for example by attaching containers of foul-smelling liquid to the raccoon, and shaving and dying portions of its fur so that it becomes black and white with a puffy tail.
Importantly, the researcher emphasizes that such operations result in the item being transformed so that it represents a different object of the same type: a living natural kind is made to look like another living natural kind.
Children are finally presented with an "after" image of the subject, and asked how they?d categorize it post-surgery?if it?s a raccoon or a skunk.
Their answers are rated on a scale of 1 to 3. In this system, if children shift their categorization?if they assert that the shaved raccoon is now a skunk, as it resembles this animal in appearance and smell?they?re given a score of 1.
In contrast, if children are unsure of how the transformed thing should be categorized?if it?s some ?ra-kunk? hybrid of the two?they?re given a score of 2.
Finally, if children don?t change the initial categorization of the item?saying that the animal remains a raccoon, despite its altered physical state?they?re assigned a score of 3.
Here, the dependent variables are the average scores children receive across natural kind and artifact stories, which can be used to gauge children?s tendency to resist changing their identification of transformed objects.
Based on research performed by Frank Keil and colleagues, it is expected that children will readily change their categorization of inanimate artifacts, but not of living natural kinds, presumably due to some internal, unchangeable feature that animals possess.
Before the experiment begins, create cards with images of pre- and post-transformation items. Design 16 such transformation vignettes?eight dealing with artifacts, and eight dealing with natural kinds.
When the child arrives, greet them and introduce them to the concepts of doctors and surgeries.
Present the 16 transformation vignettes to them in a random order: ?The doctors took a coffeepot that looked like this. They sawed off the handle, sealed the top, took off the top knob, sealed the spout, and sawed it off. They also sawed off the base and attached a flat piece of metal. They attached a little stick, cut a window in it, and filled the metal container with bird food. When they were done, it looked like this.?
?After the operation, was this a coffeepot or a birdfeeder??
If in any instance the child gives an ambiguous or hybrid response?like saying that a coffeepot transformed into a birdfeeder is now a ?coffee-feeder??also ask them to call it by either its pre- or post-transformation name. This encourages children to think about their answers, as well as warrants different responses across all of the vignettes.
Question the child in a free-form manner about the reasons behind their response, in order to determine the features or principles of the post-transformation item that led them to their conclusion.
For each vignette, record the child?s response for future analysis.
When the child has completed all 16 vignettes, have two independent raters read their responses and code them on a scale of 1?3.
To analyze the data for each of the two types of transformations?either natural kinds or artifacts?plot the average children?s score as a function of age.
Perform an analysis of variance to determine if there are differences between the three age groups or two types of transformations.
Notice that as age increases, the average scores of children for the artifact transformations remained relatively constant, hovering around 1.25. This indicates that children at all ages studied readily change their categorization of altered artifacts.
In addition, these results suggest that the perceptual features of post-transformation items in these instances?what a child sees?informs their categorical placement. In contrast, for natural kind transformations?especially those involving animals?children?s average scores increased as a function of age. This indicates that as children get older, they gradually represent category membership as an internal, unchangeable aspect of animals.
Now that you know how researchers are using transformation scenarios to better understand how children categorize general artifacts and natural kinds, let?s look at other applications of this technique.
As you?ve learned, natural kinds can be either living items?like plants or animals?or inanimate objects, like minerals.
Thus, some researchers are further looking into how children come to categorize these different types of natural kinds, and if any differences exist between these processes.
Such work has demonstrated that whereas children quickly learn to categorize animate natural kinds?and resist the tendency to change their categorization of animals post-transformation?a similar ability for inanimate natural kinds develops more slowly.
Some psychologists have suggested that parental influence may play a role in this phenomenon. For example, a parent may emphasize that an animal should be categorized as a fish if its parents were fish, and if it lays eggs that also yield fish.
However, given the advanced subject matter of what makes an inanimate natural kind?like the organization of electrons in its atoms, or the elements that constitute it?parents may not have similar discussions with their children about such items.
Other researchers are looking at how children apply facts about a single member of a category to make inferences about other items in the same category.
For example, psychologists could tell children that a mouse is more active at night and likes to eat cheese, and then show them a collection of realistic pictures composed of mice with different coats, other rodents, and unrelated animals?like a cow.
By either including or removing labels from these images, researchers can evaluate the extent to which textual category labels, in addition to morphological features, factor in children?s ability to infer that other mice have the same characteristics.
You?ve just watched JoVE?s video on how children develop the ability to categorize?and make inferences about?natural kinds and artifacts. By now, you should know how to design and use transformation stories to investigate this phenomenon, and collect and interpret results. You should also have an understanding of how psychologists are using this method to investigate other aspects of the categorization process.
Thanks for watching!
View the full transcript and gain access to JoVE Science Education videos
Q1: What is the difference between natural kinds and artifacts in child categorization research?
Natural kinds include living things like animals and plants, while artifacts are non-living, man-made objects such as cups or pots. Researchers use this distinction to study how children categorize items. Frank Keil's transformation studies revealed that children treat these categories differently, resisting changes to animal identities but readily reclassifying altered artifacts based on perceptual features.
Q2: How do transformation stories help researchers understand children's categorization abilities?
Transformation stories present children with narratives where objects or animals are physically altered to resemble something else. Researchers then ask whether children maintain the original category or shift to the new one. This technique reveals whether children rely on perceptual features or internal, unchangeable properties when categorizing items across different age groups.
Q3: What do children's responses reveal about how they categorize transformed animals?
Children's responses are scored on a scale of 1 to 3: shifting categorization scores 1, uncertain hybrid responses score 2, and maintaining original categorization scores 3. Results show that as children age, they increasingly resist changing animal categorizations despite physical transformations, suggesting they recognize internal, unchangeable features that define natural kinds.
Q4: Why do children readily change their categorization of transformed artifacts but not animals?
Children categorize artifacts based on perceptual features and function, so altered appearance leads to reclassification. Animals, however, are understood to possess internal, unchangeable properties that define their category membership. This developmental distinction reflects children's growing understanding that category membership for natural kinds depends on hidden biological features rather than observable characteristics.
Q5: How do inductive inferences relate to children's category knowledge?
Children use inductive inferences to make educated guesses about unfamiliar category members based on what they know about a single item. For example, if a child knows their dog wags its tail when happy, they infer that other tail-wagging dogs are also happy. This ability demonstrates how category knowledge enables children to extend properties from known members to new instances.
Q6: What age-related changes occur in how children categorize natural kinds versus artifacts?
Children's average scores for artifact transformations remain constant across ages around 1.25, indicating consistent reliance on perceptual features. In contrast, scores for natural kind transformations increase with age, showing that older children increasingly recognize internal, unchangeable aspects of animals. This developmental pattern suggests how children solve problems using causal reasoning about biological identity.
Q7: How do parental discussions influence children's categorization of animate versus inanimate natural kinds?
Children quickly learn to categorize animate natural kinds like animals and resist reclassifying them after transformation. However, categorization of inanimate natural kinds like minerals develops more slowly. Researchers suggest parents emphasize biological criteria for animals, such as parentage and reproduction, but may not discuss comparable internal features of inanimate objects with their children.