Week 1: Chapter 5 and 7
Chapter 5
Conceptual development and the biological world
● "Conceptual knowledge" is how we understand and organize the world around us,
including objects, actions, events, and even people's thoughts.
● Babies start building this knowledge early by noticing patterns, linking features (like
"soft" with "cat"), and figuring out causes and effects (e.g., "If I drop this toy, it falls").
● This ability helps us categorize things (group them) and make predictions (e.g., "This
soft, furry thing is a cat; other cats will also probably be soft and furry").
Categorization is the mental process of sorting things into groups:
1. Superordinate level: Very broad, like "animal."
2. Basic level: Everyday categories, like "cat" (most useful and easiest for the brain to
process).
3. Subordinate level: More specific, like "Siamese cat."
● Basic-level categories are the "sweet spot." They’re specific enough to be useful
but not so detailed that it becomes overwhelming.
● Kids also learn that some features are permanent (like wings on birds), while others
are temporary (like mud on feathers).
How does the brain help?
The brain has built-in tools for organizing sensory input into categories without us having to
think about it. For example, it automatically notices patterns and averages out features to
form "prototypes" (typical examples of a category, like a "generic bird" with wings and a
beak).
Population coding is a fancy term for how the brain uses groups of neurons to summarize
information—like figuring out the average shape, color, or size of something.
Prototype Theory
Eleanor Rosch’s prototype theory explains that categories often have "best examples"
(prototypes) that represent the group, like a sparrow being a better example of a bird than a
penguin
● These prototypes naturally emerge because many attributes (e.g., wings, beak,
feathers) tend to "bundle" together in the environment.
● Neuroscience backs this up: our brains are wired to create these prototypes quickly,
which helps us organize and understand new things
Basic-level categories, such as "cat" or "car," are foundational in conceptual development,
representing the level at which the greatest amount of information can be gained with
minimal cognitive effort. These categories are based on the co-occurrence of features in the
,natural world, such as feathers with wings in birds. Infants form prototypes—generalized
mental representations—of these categories through statistical learning, extracting feature
correlations and probabilities.
Research on Infant Categorization
Studies have shown that infants as young as three months can form categorical
representations of real-world objects. For instance, experiments by Eimas and Quinn (1994)
demonstrated that infants could distinguish between animals like horses, zebras, and
giraffes based on visual features. Similarly, Arterberry and Bornstein (2001) showed that
infants could categorize animals versus vehicles using either detailed static images or
motion-based point light displays, highlighting the brain's ability to encode higher-order
properties like motion.
Prototype Formation and Perceptual Categories
Prototype formation appears to be an automatic neural process. The brain computes
"summary statistics" of visual inputs, enabling the extraction of category prototypes even
with limited information. For example, motion alone can effectively differentiate categories
like vehicles and animals, supporting the idea that higher-level properties are processed first
(Hochstein et al., 2015).
While prototype effects initially seemed purely perceptual, researchers argue that
categorization also holds conceptual significance. Sorting behaviors in pre-verbal children,
such as grouping toy dogs with other dogs and cars with cars, suggest that these categories
are meaningful beyond sensory features. This interplay between perception and cognition
underscores the foundational role of basic-level categories in conceptual development.
Sequential touching, observed in children from around 18 months of age, involves touching
objects from the same category in sequence more often than expected by chance. This
behavior serves as a precursor to systematic sorting and provides a measure of emerging
categorization.
Studies by Mandler and Bauer (1988)
● Basic-Level and Superordinate Categories
Children were presented with toys from basic-level contrasts (e.g., dogs versus cars)
and superordinate contrasts (e.g., animals versus vehicles). Sequential touching of
toys within the same category indicated categorization.
○ Results:
■ 12- and 15-month-old infants showed categorization only at the basic
level.
■ Categorization of superordinate and contextual categories became
reliable by 20 months.
■ Categorization appeared influenced by perceptual similarity; for
example, distinguishing between dogs and cars (low similarity) was
easier than between dogs and horses (high similarity).
● Influence of Superordinate Classes:
The findings suggested that superordinate classes (e.g., animals versus vehicles)
, play a significant role in the development of categorization, as basic-level distinctions
were clearer when categories belonged to different superordinate classes.
Pauen’s Object Examination Paradigm (2002)
Pauen explored whether infant categorization relied on perceptual features alone or involved
conceptual understanding.
● Experiment:
Infants were familiarized with objects from one category (e.g., toy animals). They
were then presented with a new exemplar from the familiar category and an object
from a different category (e.g., furniture). Longer manipulation of the novel category
object indicated categorical discrimination.
● Findings:
○ Infant categorization was not solely based on perceptual similarity.
○ Infants drew on conceptual insights, even when perceptual features were
deliberately matched across categories.
○ Artificial stimuli (e.g., furniture resembling animals) were used to explore the
boundaries of categorization, showing that pre-existing conceptual knowledge
influences infant behavior.
Pauen wanted to know if infants would categorize objects better using realistic toy replicas of
animals and furniture, compared to artificial ones. If categorization depended on how similar
the objects looked, it should be harder for infants with the artificial toys, since these looked
more alike than animals or pieces of furniture did. However, if infants used prior knowledge
to group the objects, there should be no difference between the realistic and artificial toys.
Pauen found that infants spent more time looking at objects from new categories, regardless
of how similar they looked. This suggested that their categorization was based on
knowledge, not just appearance.
Key Insights:
1. Role of Perceptual Similarity:
○ Basic and superordinate categories often reflect perceptual similarities.
○ However, conceptual categorization emerges as children integrate perceptual
features with deeper structural understanding.
2. Emergence of Conceptual Understanding:
○ Infants begin categorizing based on perceptual cues but gradually incorporate
conceptual information.
○ This supports Rosch’s theory that perceptual similarity aligns with structural
similarity, providing a foundation for conceptual development.
Matching-to-Sample Task
A task where children are given a sample or target object and asked to select the correct
match from two alternatives.
Findings from Bauer and Mandler (1989)
● Experiment Design:
, ○ Triads were designed to test both basic-level (bird, nest) and
superordinate-level categories (monkey, bear, banana).
○ Children aged 19, 25, and 31 months were asked to match objects in a
“finding” game.
● Results:
○ High accuracy across all ages for both basic-level and superordinate-level
tasks, suggesting that children as young as 19 months can categorize at both
levels.
○ Basic-level matches were slightly easier than superordinate matches, but
performance remained strong for both:
■ Basic Level: 85% (19 months), 94% (25 months), 97% (31 months).
■ Superordinate Level: 91% (19 months), 81% (25 months), 93% (31
months).
Developmental Sequence: Basic vs. Superordinate Levels
Contrary to Rosch’s theory that basic-level categories develop first, researchers like Mandler
and Quinn propose a global-to-basic sequence:
● Mandler's View:
○ Sensitivity to superordinate categories may precede basic-level distinctions.
○ Infants start with general representations (e.g., animals vs. vehicles) that
refine into narrower categories over time.
○ Evidence: Studies using sequential touching and object examination
tasks suggest that infants initially distinguish global domains rather than
specific basic-level categories.
● Quinn’s View:
○ Development is driven by perceptual learning processes without requiring
conceptual understanding.
○ Computational models and experiments in visual neuroscience support a
global-to-basic developmental sequence, as infants learn to differentiate
broader categories before focusing on finer distinctions.
Challenges to Rosch’s Theory:
1. Global-to-Basic Sequence: Infants might first make broad categories (like animals
vs. vehicles) before narrowing down to more specific ones (like dogs vs. fish). This
suggests that learning broad categories helps them later make finer distinctions.
2. Conceptual Limitations of Experimental Designs: If infants fail to tell the
difference between toy dogs and toy horses, it might not mean they don't know the
difference. The problem could be with the toys or the way the task is set up.
Key Takeaways
